LSST Data Management Cybersecurity Draft Plan July 30, 2009 Bill Baker, NCSA Kirk Borne, GMU Tom Handley, IPAC Jeff Kantor, LSST Jim Hughes, NOAO Ron Lambert, NOAO Lee Le Clair, Ephibian Heather Larrieu, SLAC Ray Plante, NCSA Introduction ................................................................................................................................................................3 LSST’s Overall Goals ...........................................................................................................................................3 Fault Tolerance ....................................................................................................................................................3 Design Principles .................................................................................................................................................4 Strategies ................................................................................................................................................................4 Costs..........................................................................................................................................................................6 Scope of this Document ....................................................................................................................................7 Organization: Sites and Subsystems............................................................................................................7 Partners and Sponsoring Institutions .........................................................................................................9 To Be Defined During Construction Phase ............................................................................................ 10 1 Common Policies ......................................................................................................................................... 10 1.0 1.1 1.2 1.3 1.4 1.5 2 DM Subsystems ............................................................................................................................................ 19 2.1 2.2 2.3 2.4 2.5 2.6 2.7 3 Observatory Control Subsystem ................................................................................................ 20 Archive Operations Subsystem ................................................................................................... 22 Distributed Processing Subsystem ............................................................................................ 25 Community Service Subsystem................................................................................................... 26 Visitor Network ................................................................................................................................. 27 Event Messaging Component ...................................................................................................... 29 Operation Center .............................................................................................................................. 33 Sites ................................................................................................................................................................... 33 3.1 3.2 3.3 3.4 4 Definition of Terms .......................................................................................................................... 11 Responsibilities ................................................................................................................................. 11 Subsystems.......................................................................................................................................... 13 Science Data ........................................................................................................................................ 13 Physical Operating Environment ............................................................................................... 16 Management, Operational, and Technical Controls ........................................................... 16 Summit and Base Site (JK - Ron Lambert review for accuracy) .................................... 33 Archive Site ......................................................................................................................................... 37 Standalone Data Access Centers ................................................................................................. 38 Headquarters and Education and Public Outreach Center ............................................. 39 Threats............................................................................................................................................................. 40 Introduction........................................................................................................................................................ 40 4.1 Vulnerabilities .................................................................................................................................... 41 4.2 Control Analysis ................................................................................................................................ 42 4.3 Likelihood Determination ............................................................................................................. 43 4.4 Impact Analysis ................................................................................................................................. 44 Draft LSST Security Plan Page 3 of 49 Introduction This document presents policies and architecture to address cybersecurity across LSST Data Management facilities. LSST has significant cybersecurity demands. It will be a highly visible, premier scientific endeavor with both open data and real-time processing demands; it will attract attention from friends and troublemakers alike. At the same time, its staff and user base will be large and diffuse; they will need secure and convenient access to its infrastructure and services. LSST’s Overall Goals LSST has two apparently conflicting goals: Openness – generous access to data by scientists and the public, in particular within regions and nations that sponsor LSST, and in some cases world-wide Security – long-term data integrity and dependable infrastructure LSST’s cybersecurity priorities are to protect data, infrastructure, and software processes. 1. Protect data against corruption and loss due to unauthorized access a. Data in transit between sites, such as from Chile to the Archive Center b. Archive data, both raw and processed c. User-generated data in scientists’ personal LSST scratch space 2. Protect data processing systems—both software and hardware—that are integral to LSST’s science goals a. b. c. d. Prevent unauthorized access to infrastructure Secure communication among automatic software processes Provide secure administrative access for staff Provide appropriate security mechanisms for external users, including scientists and the general public, balancing ease of use and security e. Prevent and detect misuse of LSST resources (hijacking, malware) 3. Detect and remedy actual and attempted violations a. Constitute a cybersecurity team that is ready to handle incidents. b. Maintain monitoring systems on the computing infrastructure, network and gateways. Fault Tolerance LSST’s goal of dependability hinges on both security and fault-tolerance, the latter of which is articulated in the LSST DMS Fault Tolerance Architecture document. When a fault is detected, the responsibilities will be: Fault Tolerance: o Recover from the fault and continue operations, regardless of the origin of the fault (hardware failure, security breach, etc.) Page 4 of 49 LSST Security Plan Draft Cybersecurity: o o o Investigate whether the fault is due to a security failure Correct the security failure and consider preventative measures Follow up with an incident response and investigation if appropriate A security breach will not necessarily cause a fault directly. However, correcting it may require engaging fault tolerance measures; for example, if a server is compromised, a failover may need to be engaged while the breach is remedied. Design Principles The cybersecurity architecture is implemented as an integral part of the overall system design. Cybersecurity considerations enter into the decision process at the architectural, physical, and logical levels. These considerations are: Architectural: the architecture of the system considers cybersecurity and data integrity issues as a fundamental tenet of the design process. For example, end user access outside of observatory staff is only permitted via the Data Access Centers. Physical: the entire LSST System is protected by access filtering at the gateway(s); only selected systems are permitted full access. Critical systems are further protected by isolating them from any direct access by any outside user, system or network. For example, Alert and Data Release Production servers at Archive Center are isolated from end user accessible servers. Logical: only selected interface protocols on selected servers are open for outside access, and secure user authentication is used as broadly as possible. For example, public-facing EPO servers are firewalled for all traffic except necessary protocols such as HTTP and FTP. Strategies LSST will follow two main cybersecurity strategies: Network perimeters around major subsystems o o o o Minimal open ports and protocols Gateways for all external Internet traffic and some inter-subsystem traffic Configuration standards (approved software) for hosts within network perimeters Convenient “Visitor Network” for non-approved hosts Appropriate authentication of users and software processes o o o Two-factor authentication for administrative access Revocable cryptographic keys for software processes Adjustable authentication and authorization for external users (scientists, public) Draft LSST Security Plan Page 5 of 49 Network Perimeters around Major Subsystems LSST will set clear boundaries around the various pieces of its infrastructure, and control access at these perimeters. In the diagram below, the green boxes, representing “Systems” and “Subsystems”, mark the main security perimeters. Exposure to the public Internet is limited to: Community Service Subsystems (CSS) present LSST’s public face. These are the only systems that permit direct end user interaction. Visitor Networks (VN) provide Internet access for staff and visitors at each site and are not trusted by the rest of LSST’s infrastructure. Trusted, locked-down gateway machines in each of the internal Systems. LSST Sites and Systems. Systems and Subsystems (green) are primary security perimeter. Multiple Systems are housed together at Sites (brown outer boxes). Needed: simplified version of this diagram, but include reference to more detailed version in Infrastructure Design. Authentication of Users and Software Processes Access to all internal systems will only be allowed through designated gateway machines, and will generally require: Host-based certificates (for automatic software processes) that are subject to both revocation and expiration Page 6 of 49 o o LSST Security Plan Draft Encryption is required for control channels, preferably using the host-based certificates that are used to authenticate the connection Data channels across Subsystem boundaries must be checked for integrity, if it is not feasible to encrypt them Administrators will use two-factor authentication, such as one-time-password (OTP) hardware tokens, for administrative access to internal systems. Configuration Standards for Internal Networks Only approved configurations will be allowed to connect to internal networks. In particular, the DM Technical Control Team, in conjunction with the cybersecurity team, will approve operating systems, software configurations, and kernel revisions for hosts that operate within subsystems’ network perimeters. For convenience, a separate “Visitor Network” will be created at each site. It will be more permissive than the internal networks, but it will be topologically external to LSST, subject to the same cybersecurity precautions as the public Internet. Costs Costs to implement this cybersecurity plan are discussed below qualitatively. Quantitative costs are included in the LSST DMS MREFC and Operations cost estimates. Infrastructure Cybersecurity requirements that are likely to affect LSST infrastructure costs: Two-factor authentication system o Authentication Server. Modest processing requirements. May be shared with a partner organization such as NOAO or NCSA. o Tokens for LSST system administration staff. Current costs are around $100 per hardware token, which is typically a keychain dongle or small calculator form factor. Options will continue to proliferate over the next few years, potentially driving the price down and improving convenience. o Note that end users will not require two-factor authentication; only LSST system administrators will need them. Username/passphrases is expected to be sufficient for end users. Network partitioning o Commercial routers are capable, in their standard configurations, of implementing LSST’s network partitioning plans, at no additional cost. For example, we expect that Firewalls, VLANs, and even some packet inspection will be built into the router configuration, without the need for separate hardware. Monitoring o Traffic analysis and intrusion detection will require separate physical hardware or virtual machines at each Site or Subsystem and at the Control Center. The individual site monitoring mechanisms are embedded in a software WBS element in the DM cost estimates. See monitoring and logging, section 1.5.3. Draft LSST Security Plan Page 7 of 49 Management Cybersecurity management costs include: Testing and verifying that cybersecurity requirements are met Helping users with authentication and access difficulties Administering hardware tokens and host certificates Debugging and maintaining network configurations Refer to the Responsibilities section (1.1) below for a more detailed description of cybersecurity personnel roles and responsibilities, and to the LSST Operations Plan for staffing levels. Software Development LSST software development will be impacted by cybersecurity requirements, mainly by authentication requirements when communicating across Subsystem boundaries. Some of the open-source software that LSST intends to leverage already sufficiently supports certificate-based security, but we will have to do additional work for integration and when developing custom software. This effort is included in LSST software development cost estimates. Scope of this Document This document addresses LSST’s cybersecurity plan at an abstract level. It does not specify implementation details, except as examples to demonstrate feasibility. Even so, we believe that LSST’s cybersecurity requirements can be met at reasonable cost by current technologies and products, including network routers and firewalls, intrusion detection, and authentication. In particular, current commercial firewalls and routers are capable of meeting LSST’s traffic requirements. Organization: Sites and Subsystems This document is mostly organized by DM Subystem. LSST infrastructure is made up of selfcontained subsystems, each in their own logical network partitions, which are installed together at physical Sites, as portrayed in the following diagram. The network partitions, with carefully controlled interfaces, delineate cybersecurity and organizational boundaries. Page 8 of 49 LSST Security Plan Draft LSST Sites and Subsystems. Sites (brown outer boxes) house combinations of Subsystems (green). Cybersecurity policy is defined mainly at the level of Subsystems. Need to update diagram to use WBS terms (subsystem, component, etc). Legend: Brown: Hardware & other Infrastructure Green: Systems Blue Text: Software components, data Blue Lines: Software control Black Lines: Bulk data transport Red: Event channel communication Definitions: Site: Physical space in a building managed by an institution that hosts LSST DM Subsystems. DM Subsystem: A defined combination of configured hardware, a software stack, a set of running processes, and the people that manage them. DM Subsystems Observatory Control Subsystem (OCS): controls the camera, telescope and observatory hardware, and data acquisition. Draft LSST Security Plan Page 9 of 49 Archive Operations Subsystem (AOS): manages on-going operations of a site, including long-term storage, the main archive storage cache, data access services, and the brain of operations, data management control (DMC). Data Processing Subsystem (DPS): used to run pipelines Community Services Subsystem (CSS): hosts services and associated storage that are accessed by outside users. Visitor Network (VN): a relatively open network that allows easy network access to the Internet. LSST Sites Sites each host a combination of DM Subsystems. For example: Base Site = OCS + DPS + AOS + CSS + VN Archive Site = DPS + AOS + CSS + VN Stand-alone DAC Site = AOS + CSS + VN Partners and Sponsoring Institutions LSST shares infrastructure and operations with many scientific and academic institutions, and is sponsored by a broad base of governmental and private supporters. These partners and sponsors have their own cybersecurity requirements and arrangements that affect LSST. Funding Agencies’ Cybersecurity Requirements NSF – The National Science Foundation is still formulating its cybersecurity requirements for projects that it funds. NSF-appointed Concept Design Reviewers asked that LSST include a cybersecurity plan in the upcoming Preliminary Design Review (PDR). LSST will be working with other NSF-funded institutions, such as NCSA (see below) that have their own cybersecurity policies and practices, which both LSST and NSF can look to for examples. DOE – The Department of Energy generally has extremely strict security requirements for its projects. Since these security requirements are intended to deal with extremely sensitive national security information, such as nuclear weapons programs, we do not expect to be required to follow DOE’s security standards during operation; we will focus, instead, on NSF's requirements and on our own inherent cybersecurity priorities. This plan sets the base level for all LSST sites, if certain sites are subject by location to more stringent DOE requirements, those requirements will be applied at those sites specifically. Private sponsors – We are not aware of separate cybersecurity requirements from private sponsors, other than the need to protect researchers’ proprietary data if it is derived from, created, or stored on LSST community access resources from LSST data. Participating Institutions These institutions have helped to develop the LSST cybersecurity plan NCSA (National Center for Supercomputer Applications, Illinois) o o Experience operating large open-access computational science facilities. Will operate the LSST Archive Center. Page 10 of 49 o o o Will host LSST’s South American infrastructure—telescope, base station, and Chilean Data Access Center—and provide facilities management, including physical security. Experience with open-access astronomy with Sloan Digital Sky Survey and numerous proprietary astronomy projects. Operates flight and ground missions such as the Spitzer Space Telescope and the Palomar Transit Factory, whose infrastructure and goals have close parallels to LSST. Contributing to LSST software development. Ephibian o Contributing to LSST software development. IPAC (Infrared Processing and Analysis Center, Caltech) o Draft NOAO (National Optical Astronomy Observatory) o LSST Security Plan Security and software consultants with experience in security and operations control systems especially in the defense sector. SLAC (Stanford Linear Accelerator Center) o o Experience operating secure DOE and NSF funded science projects. Contributing to LSST software development. To Be Defined During Construction Phase LSST will define the following plans and procedures during the Construction phase. They will be effective during the Construction and Operation phases. Incident Response Plan Certain technical details, based on the cybersecurity environment at the time o o o Encryption and hash strength A specific two-factor authentication mechanism Network monitoring and intrusion detection systems Security Procedures for LSST Sites o o o o o o o o Awareness and training Audit and accountability Contingency planning Media protection Physical and environmental protection Security planning Personnel security Communication policy 1 Common Policies Each DM Subsystem in LSST’s infrastructure will have a specific set of policies associated with it according to what it supports and what it needs to protect. This section describes common policies that may be referred to from the subsystem-level cybersecurity policies. Draft LSST Security Plan Page 11 of 49 1.0 Definition of Terms 1.0.1 Cybersecurity Objectives: Confidentiality, Integrity, and Availability As a useful model, Federal Information Systems define1 the “security objectives” of confidentiality, integrity, and availability for information and information systems as: Confidentiality: “Preserving authorized restrictions on information access and disclosure, including means for protecting personal privacy and proprietary information…” A loss of confidentiality is the unauthorized disclosure of information. Integrity: “Guarding against improper information modification or destruction, and includes ensuring information non-repudiation and authenticity…” A loss of integrity is the unauthorized modification or destruction of information. Availability: “Ensuring timely and reliable access to and use of information…” A loss of availability is the disruption of access to or use of information or an information system. 1.0.2 Impact Levels: Low, Moderate, and High High: Exercise of the vulnerability would have a severe or catastrophic impact on LSST, to an extent or duration that it is unable to perform one or more of its primary functions. The failure may: Severely violate, harm, or impede LSST’s mission, reputation, or interest Result in highly costly loss of LSST’s major assets or resources Cause severe or catastrophic harm to individuals or research projects Moderate: Exercise of the vulnerability could have a serious adverse effect; a system may still be able to perform its primary functions, but its effectiveness could be significantly reduced. A failure may: Violate, harm, or impede LSST’s mission, reputation, or interest Result in the costly loss of LSST’s assets or resources Cause significant harm to individuals or research projects Low: Exercise of the vulnerability could have a significant but limited effect; a system would continue to perform its primary functions, but the failure may: Reduce a system’s performance or effectiveness Cause minor damage to assets Cause minor, recoverable harm to individuals or research projects 1.1 Responsibilities Each site is expected to have one person, a cybersecurity officer, who is responsible for overall managing and cybersecurity monitoring for the site. This person will be responsible for ensuring proper security configuration of machines and processes at the local site, FITS PUB 199, Standards for Security Categorization of Federal Information and Information Systems, Computer Security Division, Information Technology Laboratory, NIST. 1 Page 12 of 49 LSST Security Plan Draft responding to alerts from cybersecurity monitoring mechanisms, and serving as the final point of contact on cybersecurity issues requiring action by a human. The Archive Site will host personnel to not only carry out these activities for the Archive Site but also coordinate cybersecurity management across all LSST sites. Associated with the cybersecurity officer is an appropriate budget per the LSST Operations Plan. In addition to local security personnel, each site will have some form of general operations staff. At the Archive and Base Sites, these will be site operators who provide 24-7 operations support. Data Access Centers (DAC) sites may rely on local site staff (who are not necessarily LSST employees) to serve in this role. As part of their routine operations, this staff will monitor the site's systems for failures and warnings. In particular, alerts from cybersecurity monitoring mechanisms will be routed first to this staff. They will make the initial assessments of failures or other unusual events that might indicate a cybersecurity problem. When a problem is suspected, they have the ability to shut down or restart processes and Systems according to a defined procedure (to be defined later) in accordance with overall Operations and Security policy. In such cases, the local cybersecurity officer would be contacted. In all cases, the procedures for reacting to possible cybersecurity issues and the use of personnel to manage site security must be consistent with the policies and practices of the local institution. In particular, the security officer may be required to report to and coordinate with an overall institutional security officer. Cybersecurity incidents will have to be submitted to the local incident reporting process as well as the LSST operations and control system. 1.1.1 Incident Response LSST’s priorities in the face of cybersecurity incidents are: 1. Recover from the incident and resume operations 2. Prevent recurrence 3. Pursue and prosecute serious or repeat attackers Even though LSST’s top priority is recovery, LSST personnel should make reasonable efforts to preserve forensic evidence, especially in the case of more serious incidents, for use in follow-up investigations. 1.1.1.1 Incident Response and Security Team LSST will designate, from among its operating staff, an Incident Response and Security Team (IRST). It should include: Individuals with system-level access to LSST’s computing infrastructure Individuals with expertise in cybersecurity incident analysis and response LSST’s security officer(s) The incident response team should maintain a 24-hour contact point that is well known to LSST staff. It would be expedient to have the contact point be the same as a general LSST help desk. JK – What is relationship of IRST to security officer, i.e. does security officer lead IRST? Draft 1.1.1.2 LSST Security Plan Page 13 of 49 Reporting Incidents Any staff member who becomes aware of or suspects a cybersecurity incident should report it to the IRST. The IRST will investigate the incident, notify affected parties (if needed), and recommend corrective actions. 1.2 Subsystems 1.2.1 Network Architecture Each major subsystem should be logically isolated—for example on its own sub-network or VLAN—to allow router-level enforcement of security policies. Communication with machines in other major subsystems should be through a DMZ where possible. Figure 1.1 Basic network perimeter for an individual DM Subsystem (need to updated diagram text to match document text) 1.3 Science Data LSST data products are divided into three groups, as described in the DM System Requirements: Level 1 data products: Automatically generated by processing the stream of data from the camera system during normal observing. Level 1 data products are therefore being continuously generated and / or updated every observing night. Page 14 of 49 LSST Security Plan Draft Level 2 data products: Generated from Level 1 as part of an annual Data Release. They include data products for which extensive computation is required, often because they combine information from many exposures. Level 3 data products: Derived from Level 1 and/or Level 2 to support particular science goals. Typically, a research team will create them, with the help of DMS infrastructure, and will store them temporarily at LSST Data Access Centers (DACs). From time to time, Level 3 products that are found to be especially useful may be archived and supported by LSST and even promoted to Level 2. Level 1 and Level 2 data products that have passed quality control tests are required to be accessible to the public without restriction. The access policies for Level 3 data products will be product- and source-specific, and in some cases will be proprietary. One of the main objectives of the cybersecurity plan is to ensure the integrity and availability of data for scientific users, both in storage and during network transfers. In this section, we highlight the classes of data that are critical both for end-user science and for the internal generation of end-user data products. Our general security concerns for the data include: Raw observational data, including its supporting metadata such as calibration and pointing, must be protected against corruption and deletion, as it cannot be regenerated. Generated data products, while they may be recomputed from the raw data, may also be very valuable, as a massive loss of generated data would have a major impact on data availability: not only would it take considerable time to regenerate products, the regeneration would take resources away from normal operations and the creation of the next generation of products. Delivery of data to users must ensure integrity and appropriate confidentiality. It should contain the intended scientific content of the project and must not inadvertently provide a vector data that is not a part of the LSST mission (malware, advertisements, illicit media, etc.). New scientific data sets created by end users from LSST products will form the basis of new publishable results; thus, these datasets (which may be stored temporarily on LSST servers) are sensitive. Even database queries they submit will carry scientific insight. A scientist must be guaranteed, while using LSST resources, that their data and session/command history are not accessible by other users apart from those with whom they explicitly wish to share them. We note again that the scope of this document relates only to the protection of data from corruption and loss by unauthorized access. Corruption and loss due to hardware failure, for example, is an issue for LSST’s fault tolerance design. Mechanisms for detecting corruption, whatever the cause, are also primarily a concern of fault tolerance, and not cybersecurity. 1.3.1 Processed Data Products Intended for Public Release By and large, external audiences will be interested in Level 2 and 3 data products—rather than the more raw Level 1 data products. Confidentiality: low. The challenge lies in making meaningful queries and managing bandwidth—not in protecting confidentiality. Draft LSST Security Plan Page 15 of 49 Integrity: moderate. We should protect integrity, but if it is compromised it can be recomputed it from archived raw data. However, detection of corruption should be regarded as a high priority. Availability: moderate. Temporary unavailability should be avoided, but is not catastrophic. 1.3.2 Archival Science Data Applications that manage archival data must be written to allow only reading and adding to archived raw data, and not over-writing. Deletion of archival data must: Only be possible with two-factor authentication, by a highly privileged user Be accompanied by warnings to the user Display progress verbosely and be interruptible Be logged, so that deleting can be tracked and reasons noted Confidentiality: low. Archival science data is publicly available; the challenge lies in making meaningful queries against it and managing bandwidth, not in protecting its confidentiality. Integrity: high. In order to be useful for science, the data's integrity, including its metadata and provenance, must be trusted. Once lost, it is not retrievable. Availability: moderate. Temporary unavailability should be avoided but is not catastrophic. 1.3.3 Internal Science Data Certain data is not intended for public release—in particular, data whose quality has not been assessed. Confidentiality: low. An individual leak of internal science data would only cause unreduced or low-quality data to be distributed publicly, prematurely, and would not significantly harm LSST's operations. It might have an impact on LSST's reputation. If it happened routinely, it would cause significant irritation and mistrust in the astronomy community. Integrity: high. Most internal science data will eventually be processed, archived, and released, and therefore its requirement for integrity is the same as Archival Science Data, above. Availability: moderate. Temporary unavailability should be avoided but is not catastrophic. 1.3.4 User-generated Data LSST will provide private storage space for researchers that is: Externally accessible via standard protocols such as VOSpace Local to LSST compute resources for efficient HPC access (Potentially) subject to spatial quotas and time limits Confidentiality: moderate. Researchers require confidentiality before publishing. Page 16 of 49 LSST Security Plan Draft Integrity: moderate. We will attempt to preserve user-generated data. If it is lost, we assume that researchers can regenerate it, even though it may be time-consuming and inconvenient. Researchers should keep backup copies of critical algorithms and code. Availability: moderate. Temporary unavailability should be avoided but is not catastrophic. 1.4 Physical Operating Environment LSST DMS will operate almost entirely on the premises of partner institutions. As such, each Site has its own unique physical environment, as described in its own section of this document. 1.5 Management, Operational, and Technical Controls 1.5.1 Access Control 1.5.1.1 Internal User Access This subsection describes how a user (as opposed to a service) who is logged into a machine inside a particular DM Subsystem may gain access to services or shells on other machines within that Subsystem. 1.5.1.1.1 Intra-Subsystem Shell Access Unless the Subsystem security policy mandates additional restrictions, shell-level access to a remote machine will be allowed only through a secure remote shell protocol that includes: An encrypted connection Client and server host authentication Cryptographically secure user authentication (SSH and Kerberos are both acceptable technologies for this.) Password-less authentication for accessing a remote shell is discouraged unless it is needed for effective use of an application (e.g. SVN); this application can only use password-less authentication within the confines of the Subsystem. Rationale: Use of SSH/Kerberos within a Subsystem can slow the propagation of unauthorized access and protect against network sniffers. 1.5.1.2 Remote User Access This subsection describes how a user (as opposed to a service) who is logged into a machine located outside of a particular Subsystem may gain access to services or shells on machines within the Subsystem. All such access shall be included in system logs described in Monitoring and Logging, section 1.5.3. Draft LSST Security Plan 1.5.1.2.1 Page 17 of 49 Public & Science User Access All access by the general public and general science user (those without administrative access) will be limited to the Community Service Subsystem (CSS), generally via a web portal or rich client interface. Refer to the specific CSS policies for details. Rationale: By definition, non-CSS Subsystems will not have any services that require direct access by public/science users. Their services can be provided to users by using the CSS as a proxy. 1.5.1.2.2 Portal-based Access A software component may support a portal (web) interface. When it does, it must support authentication and authorization policies from one or more of the four categories: 1. Passive monitoring o o The user connects only for retrieving (read-only) data. Authenticate with a username and password 2. Active monitoring o o User can change logging levels in a manner that may affect other users Two-factor (2F)/one-time password (OTP) authentication required 3. Non-critical control o o o o Non-critical generally refers to control over operations that are not dangerous to life or property User can change state of component, launch other components, etc. Two-factor (2F)/one-time password (OTP) authentication required Portal or underlying application need to determine if user has authorization for action 4. Critical control o o o o 1.5.1.2.3 Critical generally refers to control over operations that are dangerous to life or property; mainly applies to the LSST Observatory Control Subsystem (OCS) and the Telescope and Camera Control Systems (TCS and CCS) User can change state of component, launch other components, etc. In the case of OCS, this may mean controlling camera or observatory hardware. Requires authorization enabled by human on a per-session basis Two-factor (2F)/one-time password (OTP) authentication required after authorization is enabled Direct Remote Shell Access Remote shell access allowed with features required in Intra-Subsystem Shell Access, plus: Support of two-factor authentication (password-less not allowed) Remote shell access from outside a Subsystem should be restricted to designated machines within that Subsystem (e.g. login servers); it should not be enabled by default. Page 18 of 49 1.5.1.2.4 LSST Security Plan Draft Rich Client Access A rich client is an application that runs on the user's "local" machine and establishes a network connection with a specialized server application on the remote machine. Examples include 3rd party products (e.g. LabView) and custom products talking to our own services. Such a client may be used under one of the following conditions: 1. It supports the LSST two-factor authentication mechanism, and all communication with the server is encrypted. 2. It is run via an encrypted remote desktop (e.g. VNC)—the client application is actually running on an internal LSST Subsystem, with a user interface outside. 1.5.1.3 Service Authentication 1.5.1.3.1 Intra-Subsystem Software processes that communicate with each other within the protected boundaries of a Subsystem must use some form of authentication, such as host certificates or username and password. Heavy traffic flows, such as bulk data transfers, do not need to be encrypted. However, control information, such as within the Event Messaging Subsystem, shall be encrypted. 1.5.1.3.2 Intra-Site, Inter-Subsystem Software processes that communicate across Subsystem boundaries but within a single Site, for example database queries from the CSS to the AOS: Shall flow through designated gateways Shall authenticate securely, using LSST-issued host certificates where possible Shall use an encrypted channel for control signals such as commands and queries Shall encrypt data where computationally reasonable May send high-volume open data unencrypted, but shall include strong integrity checks such as checksum hashes 1.5.1.3.3 Inter-Site Software processes that communicate between LSST Sites shall follow the policy for intrasite communication, with the addition that any potentially sensitive data (such as usergenerated data) shall be encrypted. 1.5.1.3.4 Public Internet LSST software processes that communicate with the public Internet, for example, externally accessible web services: Shall flow through designated public gateway machines in the CSS Shall allow for basic authentication (username and password) on behalf of an LSST registered user Where possible, shall establish encrypted channels May, with certification by the cybersecurity officer, comply with current standards, for the sake of compatibility—for example, if a VO-compliant query mechanism that LSST supports doesn’t encrypt returned data Draft LSST Security Plan Page 19 of 49 1.5.2 Host Requirements Before a machine can be connected to an internal LSST network—other than the Visitor Network—it must be reviewed and validated by the cybersecurity officer as complying with that LSST subsystem’s security policies. Generally this will require: An approved operating system and patching strategy Cleanliness attestation from a trusted party, such as an LSST administrator, who has expertise with the machine’s operating system and installed software It may also require: An up-to-date host certificate generated by an LSST trusted Certificate Authority (CA) 1.5.3 Monitoring and Logging Logging and Network monitoring systems shall feed information in real-time to both local security personnel and to a central LSST operations center. Selected DM Subsystem components shall relay system logs and access logs into a centralized system that will store them and flag anomalies. The DM components selected to relay logs should include control machines, any machines that have network connections outside of their subsystems, and any components selected by security officers or operating staff as important to monitor. Networks shall be monitored by intrusion detection and traffic analysis systems, with particular attention to traffic across subsystem boundaries. Specific network-monitoring technologies will be determined during the construction phase. 2 DM Subsystems Jeff writes: [This section needs to be re-written to talk specifically about the security considerations in these subsystems and not about what the subsystems are and what they do, that will be covered in other documents. We should presume that the functions and data flows of each system are defined in the Infrastructure and Middleware design and operations documents and do not need repeating here. You can simply refer other users to those designs, the document will still stand alone enough. It appears that some items are characterized by Confidentiality, Integrity, Availability (CIA) assessment but not others. This should be consistent, that is, the Intro and Mgt/Ctrl sections do not include the CIA, but the Systems and Data Products sections do. Alternatively, all the CIAs could be collected in a table in an Appendix. Similarly, some Mgt/Control sections have Risk Assessments and some don’t. This should first be covered with a Common Policy specifying at least an annual risk assessment, and then subsystem-specific risk concerns can be addressed in each Subsystem section.] Page 20 of 49 LSST Security Plan Draft 2.1 Observatory Control Subsystem The Observatory Control Subsystem (OCS) is a Telescope and Site Subsystem, and not a DM subsystem, although it is included in the DM cybersecurity plan because of its close links with the DMS. The OCS has some key differences from the DMS: Much of its operation has hard real-time requirements because of its strict data acquisition cadence. It spans two physical locations: the mountaintop observatory and base facility, connected by a 30km dedicated fiber-optic link. It is not replicated—unlike DMS subsystems, there is only one OCS. It is mostly the domain of the camera and telescope teams, less so the data management team—particularly in the area of real-time requirements. Therefore, compared to the DMS, interruptions or damage to OCS systems have the potential to be both more difficult to repair and more likely to halt critical LSST operations such as capturing science data and issuing real-time alerts. Thus, security policy in the OCS is more stringent. Figure 2.1 Summit and Base Sites, including Observatory Control Subsystem (OCS) To Do: update diagram language 2.1.1 OCS Security Objectives Security objectives for the OCS are described for each subsystem separately, below. 2.1.2 OCS Responsibilities The LSST observatory operations staff will administer the OCS. The Data Management team will administer the Event Broker and Data Replicator. 2.1.3 Physical Operating Environment The OCS shares its physical environment with its site, the Summit and Base Facility—see Section 3.1.2. Draft LSST Security Plan Page 21 of 49 2.1.4 OCS Subsystems 2.1.4.1 Camera & Telescope Control Systems These subordinate control systems to the OCS are described as part of the Camera and Telescope designs. Refer to [Get document reference from German Schumacher]. 2.1.4.2 Interface with DMS Control and Management Systems 2.1.4.2.1 Data Replicator The OCS contains a real-time facility database, into which the TCS and CCS continuously feed the status of LSST’s observatory systems. The Base Facility will contain a replicator that will archive facility events and state transitions at its AOS. Confidentiality: low. Although it is proprietary, disclosure of data about the real-time state of the OCS would not cause harm to LSST. [Is that true? Could someone use the data for nefarious purposes?] Availability: moderate. Loss of the real-time facility database feed would likely halt realtime processing of science data, since the feed provides important metadata. [True?] Integrity: high. Since the real-time feed is used to generate science metadata, undetected corruption would bleed into the science data. [True?] [This needs to be discussed with German Schumacher/Dave Mills to confirm the interface and documented in the OCS – DM ICD. The replicator will likely be a client of the OCS that is provided by the OCS team and deployed as a part of the AOS. The AOS accesses the facility database using that client and does the replication. The database will not be sent to all DM facilities, but will go to the Base and Archive.] 2.1.4.2.2 Event Broker The DMS will include an Event Broker which will interface with the OCS real-time event system bidirectionally—both sending and receiving events. See Section 2.6, the Event Messaging Component. 2.1.5 OCS Science Data (Check with Gregory on how to word this section) OCS Science data includes Calibration Data as well as Raw and Crosstalk-corrected images and associated metadata. The data originates with the primary and auxiliary telescopes, external sensors, and the camera. It is transferred to the DMS via one of two paths: Primary telescope images are transferred by the CCS Science Data Acquisition Subsystem (SDS) in both raw and crosstalk-corrected forms. Calibration data and image metadata are written by the TCS and CCS to the OCS Engineering and Facility Database and subsequently accessed from there by the DMS. Page 22 of 49 2.1.5.1 LSST Security Plan Draft The above data are all collected and stored by the Base Facility AOS and also transferred to the Archive Center AOS. Calibration Data Calibration images and data are an input for all image processing, and are necessary in order to extract scientifically meaningful data. Confidentiality: low. Leaked calibration data would not be significant problem for LSST. Integrity: high. Data integrity is central to the success of LSST’s scientific mission. Availability: low. Down time is to be avoided, but is not catastrophic. 2.1.5.2 Astronomical Image Data and Metadata The CCS SDS will stream data from the telescope to the Base Site’s Subsystems. It will consist of raw images, crosstalk-corrected images, and metadata from the telescope and camera. Confidentiality: low. LSST should avoid leaking raw and crosstalk-corrected data, since it has not been examined for scientific validity, but it would not be serious problem. Integrity: high. Data integrity is central to the success of LSST’s scientific mission. Availability: moderate. Although brief interruptions are recoverable, it may be difficult to catch up after prolonged disruptions, since the LSST DM System is architected to operate near maximum capacity. See LSST Fault Tolerance for more details. 2.1.6 OCS Management, Operational, and Technical Controls Aside from the details listed below, see Common Policies, section 1.5. 2.1.6.1 Access Control Access control policies for the OCS will be determined by the Camera and Telescope teams. The OCS will use an identity and authentication system provided by Data Management. 2.2 Archive Operations Subsystem Every LSST DM Site (apart from the summit) includes an Archive Operations Subsystem (AOS). The purpose of this system is to host and manage the on-going operational processes of the site. The AOS hosts the most important assets of a site. It directly manages the data that is served to outside users as well as to internal processing within the Distributed Processing Subsystem. The long-term storage subsystem holds permanent copies of released data products that would be accessed not only for data reprocessing, but also for disaster recovery. Consequently, this Subsystem will feature stronger protections against unauthorized access than other systems, as well as the strictest integrity guarantees. Draft LSST Security Plan Page 23 of 49 2.2.1 AOS Security Objectives Unless otherwise specified, refer to Common Policies on Science Data, Section 1.2. Each type of data has different security priorities, but most AOS systems will need to support all types, so their security goals are the most-restrictive combination: Confidentiality: moderate. Limited by user-generated data, Section 1.2.4. Integrity: high. Limited by Archival and Internal Science Data, Sections 1.2.2 and 1.2.3. Availability: moderate. Shared by all science data. 2.2.2 AOS Responsibilities In addition to the site-wide security personnel (see section 2.1), the AOS will have associated with it personnel who will be responsible for operating and monitoring the dayto-day processes of the site. These personnel will handle the typical faults and may be the ones to discover evidence of a security problem. These personnel must be aware of the procedures for reporting and reacting to possible security issues. 2.2.3 AOS Physical Operating Environment The AOS shares physical facilities with its Site. See Common Policies—Physical Operating Environment, Section 1.3. 2.2.4 AOS Components 2.2.4.1 Long-term Storage Component Not all sites will have this component. This component provides permanent storage for core data products, including raw images, calibration data, release catalogs, and co-added sky images. 2.2.4.2 Data Cache Component This is component provides temporary storage for the datasets of greatest interest at any given time. It is assumed that any data stored here can be automatically reproduced from another source, either by replicating a copy from another site or recreating the dataset from original data in long-term storage. Interfaces: The AOS provides a read-only interface to the data cache component to the DPS and CSS. 2.2.4.3 Database servers These servers provide access to the data catalogs, both the previously released catalogs (which are static) as well as the current unreleased catalog (if available at the site). Interfaces: The AOS provides a read-only interface to these servers for use by the CSS and the DPS. If the site features a DPS for internal production, the AOS will provide a read-write interface exclusively for updating the current unreleased catalog. Page 24 of 49 2.2.4.4 LSST Security Plan Draft Dataset Access servers These servers provide access to file-based data products. Data access services will first look in the Data Cache for requested data sets; if the data products are not cached, the service may request them from external sites or trigger their re-creation (via the DPS) and recaching in the Data Cache. Interfaces: The AOS provides a read-only interface to the CSS and outward-facing DPS, and a read-write interface to an internal DPS, if one exists. 2.2.4.5 Data Replicator service This service copies datasets from and to other sites, via secure, authenticated connections. 2.2.4.6 Event Broker The Event Broker routes Event messages in and out of the AOS. When exchanging events with other systems either on site or off, it is done via secure, authenticated connections to other Event Brokers on defined ports. This component will also maintain its own historical database of events. For more details, refer to section 2.6, Event Messaging Component. 2.2.4.7 Miscellaneous Operational Processes The AOS will host all of the processes that carry out automated operations of data management. This includes the routine execution of pipelines on the DPS, and packaging and distributing science Alerts. See section 1.3.3.3, Service Authentication, for security requirements. 2.2.4.8 Administrative Portal Operational processes will be launched, controlled, and monitored from outside the system (possibly from off-site) via an Administration Portal. Only LSST operations personnel shall have access to this interface. It shall conform to section 1.3.3.2, Remote User Access. Confidentiality: high. Since the administrative portal can control operational processes, unauthorized access could allow interference with LSST’s core functioning. Integrity: moderate. Loss of integrity at the administrative portal would be an inconvenience, but it would not prevent LSST’s operation. As a backup to the portal, remote personnel can contact on-site operators by other means. Availability: moderate. As a backup to the portal, remote personnel can contact on-site operators by other means. 2.2.5 AOS Science Data The AOS provides the primary storage for the following products: Raw Science Data (crosstalk-corrected data is not stored) Draft LSST Security Plan Page 25 of 49 Temporary Processed Data Products Released Data Products Unreleased Science Data User Science Data See section 1.2, Common Policies for Science Data, for security considerations for these types of data. 2.3 Distributed Processing Subsystem Distributed Process Subsystems (DPSs) perform internal processing for LSST: Real-time processing (at the Base Site) Nightly processing (Archive Site) Processing for annual data release (Archive Site) A DPS is a purely internal DM subsystem with no major public or cross-site network connections. As such, its administration and security arrangements are simpler than other DMS subsystems. 2.3.1 DPS Security Objectives Confidentiality: low. A DPS is an internal DM subsystem that processes only Level 1 and 2 science data (see Science Data, section 1.3), which are intended for eventual public release. A premature release would not cause major problems, as long as it is not construed to be an “official” LSST release, which could damage LSST’s reputation for data quality. Integrity: moderate. The computations performed by a DPS can only be relied upon if it is known to not be compromised. Availability: moderate. Individual nodes within a DPS are expected to have low availability requirements, and systems that use them must be tolerant of faults such as disk failures and memory errors, but the DPS as a whole must be generally available for LSST use and is therefore rated moderate. 2.3.2 DPS Responsibilities A DPS is always associated with an AOS; DPSs will be operated and maintained by the same personnel who operate the associated AOS. 2.3.3 DPS Physical Operating Environment A DPS will be housed next to its AOS and thus will share its physical environment. 2.3.4 DPS Components TBD 2.3.5 DPS Science Data The DPS will host LSST public data (Levels 1 & 2). For security objectives, see common policies on Science Data, section 1.3. Page 26 of 49 LSST Security Plan Draft 2.4 Community Service Subsystem The Community Services Subsystem (CSS) is the community’s portal into LSST data and services. It provides: The community access web portal Service-based interfaces, such as VOSpace, for researchers Private workspaces and compute resources for researchers As the data server to the community, the CSS balances user convenience, fairness, and security. It operates in a distributed heterogeneous environment that is connected to the Internet to serve LSST project installations and users around the world. Significantly, the CSS shall provide fair real-time allocation of bandwidth and compute power, based on user profiles, which requires a robust user identity system. 2.4.1 CSS Security Objectives The CSS provides access to all types of LSST data products. Its security objectives are driven mainly by its handling of researchers’ private data, as described in Section 1.2.4. Confidentiality: moderate. Researchers’ private data, code, and research methods must be protected. Integrity: moderate. Users need to be able to trust the integrity of their private research data and public LSST data that they retrieve through the CSS. Availability: moderate. 2.4.2 CSS Responsibilities The CSS shall be administered together with the other Subsystems in its Site. For details, see AOS Responsibilities, section 2.2.2. 2.4.3 CSS Physical Operating Environment The CSS shares physical facilities with its Site. See Common Policies—Physical Operating Environment, Section 1.3. 2.4.4 CSS Components The “Sites and Subsystems” diagram illustrates when CSS fits within each Site’s overall environment including high-level services provided. The CSS provide general access to the public services of LSST. 2.4.5 CSS Science Data The CSS provides data discovery and delivery services to the LSST user community. As such, all publicly released data and their associated metadata are available through the CSS. In addition, it provides resources to further process and store tertiary scientific data in private scratch space that it provides to researchers. Draft LSST Security Plan Page 27 of 49 The CSS provides access to the following data whose permanent storage is in the AOS, but which may be cached and post-processed on the fly within the CSS: Raw observational data and processed Level 1 data products (e.g. Alerts, and DIAsources) (“Level 1” as described in Section 1.3) Processed Level 2 data products (e.g. Source and Object catalogs) (“Level 2” as described in Section 1.3) In addition, the CSS provides primary storage and compute facilities for: Researchers’ private data, derived from public LSST data (“Level 3” as described in Section 1.2), which must be kept private to protect researchers’ work Researchers’ metadata, code, queries, which also must be kept private The main security drivers are fairness and isolation. Fairness: Applications running in the CSS will need to allocate bandwidth and computation resources fairly based on users’ privileges. Isolation: All users have access to all LSST data, but the data they generate using LSST facilities is private and protected from other users. 2.5 Visitor Network Visitor Networks (VN) at each LSST site provide general-purpose connectivity to the Internet. These networks are logically located outside of the core science networks and will be regarded as part of the open Internet. The LSST sites expect to receive large numbers of visitors, including researchers, conference attendees, contractors and other personnel, most of whom need access to Internet-based services. To meet the needs of this diverse community LSST project sites will implement a Visitor Network. This network is logically located outside of the core networks. It will be primarily a wireless network, but will have some hardwired Ethernet connectivity. The goal of the network is to provide Internet connectivity for visitors and non-secure machines. Users that require connectivity to internal resources must use one of the remote access methods provided by LSST; there is no direct connection between the Visitor Networks and internal LSST project networks. All traffic from the Visitor Networks will be treated the same as traffic from the open Internet. Host institutions may already have visitor networks that can serve in the place of a network provided by LSST. 2.5.1 VN Security Objectives Confidentiality: low. The Visitor Network, like the public Internet provides no confidentiality guarantees. Integrity: low. Availability: low. The Visitor Network will make a reasonable attempt to be available at all times, but it is not a core LSST service. Page 28 of 49 LSST Security Plan Draft 2.5.2 VN Responsibilities The Subsystem Owners for individual implementations of the Visitor Network are designated for each site in the LSST Operations Plan. 2.5.3 VN Physical Operating Environment The Visitor Network resides within the physical environment described for each of the three security domains the Base Site at La Serena, the Archive Site at NCSA, and the standalone DAC sites (if any). The security polices and controls relevant to the servers and network devices that serve as the core infrastructure for the Visitor Network are covered in the LSST Network Design (document-7354). 2.5.4 VN Subsystems The “Sites and Subsystems” diagram shows how the Visitor networks fit within the site’s environments. The Visitor networks provide general-purpose access to the Internet. This requires a different security model than the other LSST general support systems. 2.5.4.1 Wireless Access Points The Wireless Access Points (WAPs) provide Internet access for visitors to the LSST sites. Anyone physically located on the site with a system that has wireless capabilities can access the wireless networks. The WAPs’ security includes controls that facilitate what the reasonably expected level of service is for users of the Visitor Networks. Policy does not allow anyone other than designated infrastructure managers at the individual sites to install WAPs. Each site shall have the capability to detect and removed unauthorized WAPs. The signal from wireless access points shall be generally contained within the physical boundaries of the sites. Visitor network users shall only be able to connect to LSST internal resources via approved externally facing services. They shall be treated the same as any general entity on the Internet. Upon connection to the Visitor Network, users shall be presented with a personal user agreement that identifies the network and presents the acceptable use policy and the conditions of use policy. Acceptance of the appropriate policies shall be required prior to allowing full access to the Visitor Network services. 2.5.4.2 Ethernet Ports Although the primary method of connection to the Visitor Network will be wireless, there will be Ethernet ports provided. Security controls on these hardwired ports shall be equivalent to those on the wireless network. When users initially access the network they shall be presented with the same policies as on the wireless network. As in the wireless network, there shall be no direct access between the Visitor Network wired infrastructure and the LSST project’s internal networks. Draft LSST Security Plan Page 29 of 49 2.5.5 VN Science Data The Visitor Network will not bear any special responsibility for LSST data products. 2.5.6 VN Management, Operational, and Technical Controls 2.5.6.1 Access Control Unlike other LSST networks, the visitor networks offer open access to the Internet. Specific access controls for this network shall include: Wireless Access points The SSID shall not be broadcast or shall have an SSID indicative of their nature (e.g. Visitor, Guest) Ethernet Backbone Users will not be required to authenticate, but must agree to the acceptable use policy prior to being granted access. Infrastructure systems Administrator access to the servers in these networks shall be consistent with central networks management policies. The Visitor Network will not require authentication to use, unless host institutions’ policies say otherwise. If authentication is required, it may be through the host institution rather than LSST. 2.5.6.2 Awareness and Training Visitors to the sites who use the Visitor Network for Internet access only and who do not have internal accounts may receive no training other than acknowledgement of the Acceptable Use Policy, but this should include information about the security of the network and indicate the level of monitoring. 2.6 Event Messaging Component The Events component functions as a nervous system for the LSST infrastructure. It is a general-purpose publish-subscribe message system based on ActiveMQ. Jeff says to move (copy?) the rest of this summary to the Infrastructure Reference Design: Some of its primary uses are: Logging Pipeline coordination VOEvent alerts Internal status notifications Error & exception alerts Each LSST DM Subsystem has a single Event Broker that is the sole source and sink of messages within that subsystem. Broker policies determine which events are routed intersubsystem, and which remain local. Page 30 of 49 LSST Security Plan Draft Event messages are transmitted over transport-layer security and include information about their origin (host and timestamp). Since they are unsigned, the EMC carefully controls their creation and takes measures to prevent false message injection, including authenticating brokers and disallowing “upstream” transmission (for example from a DAC to the OCS). Figure 2.2 The Event component, spanning multiple LSST Sites and Subsystems, with signed channels (red) among brokers and event producers. To Do: update diagram language to match document (subsystem, component, etc.) 2.6.1 EMC Security Objectives Confidentiality: moderate. Some events may be intended for a limited audience, but leakage is not considered to be a critical failure. In particular, events may contain researchers’ confidential personal data and methods. Security information such as passwords and private keys should never be transmitted via events without additional safeguards such as out of band encryption. Integrity: high. The highest priority is accuracy and authenticity of events. Certain types of events must only originate from certain systems—for example, only the real-time pipeline at the Base facility may publish Moving Object notifications. Injections must be prevented. Availability: moderate to high. The EMC availability goal is significantly higher than the DMS as a whole. With reasonable measures such as redundant hardware and automatic failover, we should be able to achieve high-two-nines reliability (such as 99.5%, independent of externalities that would also affect the subsystems in which the Event Brokers are embedded). For more in-depth discussion, see the LSST Fault Tolerance design document. Draft LSST Security Plan Page 31 of 49 2.6.2 EMC Responsibilities The EMC backbone shall be centrally managed, to ensure uniform compatibility and security policy enforcement. Individual event brokers and event sub-networks may be managed centrally, or their management may be delegated to administrators of the Systems where they reside, according to centrally-defined policies. All event brokers shall adhere to LSST policies, and LSST will provide software and guidelines for maintaining them. 2.6.3 EMC Physical Operating Environment The components of the Event subsystem will share the physical environment of each subsystem where they reside. Event Brokers will be virtual machines that share hardware with other control components. 2.6.4 EMC Components 2.6.4.1 Event Brokers The core of the EMC is a tight confederation of autonomous brokers. Each broker has its own policies that determine how events are propagated: Import: from outside the local Subsystem (from other brokers) into the broker's subsystem Export: exposing events from inside the local Subsystem to external brokers Local: events that originate locally and are not offered to external brokers Brokers authenticate to each other using centrally-issued host certificates that are revocable and which expire and must be re-issued after a set time (probably a year). If a broker presents a revoked certificate, other brokers will refuse to connect to it. A recently expired (or soon-to-expire) certificate shall provoke a warning. Brokers authenticate with their local clients (event consumers) using a host certificate as well. It may be the same host certificate used to connect with other brokers or a locally issued one. 2.6.4.2 Event Producers Event Brokers are responsible for verifying and propagating events that originate within their local subsystems. To export events to other Brokers, they shall adhere to the policies described here or risk having their host certificates revoked. Event producers should authenticate with their local broker using a host certificate. If the events are to be exported outside of the local subsystem, the producer’s host certificate has the same requirements as event brokers’ certificates. 2.6.4.3 Event Consumers Any application in the subsystem is permitted to subscribe to (consume) events. Event consumers subscribe to events from their local broker but are not required to present host certificates. They shall connect via a secure channel and authenticate their broker's identity based on its host certificate. Page 32 of 49 LSST Security Plan Draft 2.6.5 EMC Data The EMC manages the flow of control data, rather than bulk observational data. The main concern is with integrity and authenticity—event messages must really be from whom they claim to be from. 2.6.6 EMC Management, Operational, and Technical Controls 2.6.6.1 Monitoring and Logging Messages shall be logged, with 2.6.6.1.1 Detecting Anomalies Messages shall be monitored and filtered for anomalies, in an effort to detect interruptions as well as false messages such as injections and malfunctions. The central system log monitor may be sufficient, or a separate system may be needed. More details will be determined during construction. 2.6.6.1.2 Event Archive The EMC will maintain a central database at the AOS that archives events from across the DM system. If it is necessary to periodically flush low-interest messages, those of potentially long-term interest shall be retained. Specific policies shall be developed once the contents of messages are better understood, and shall be reviewed regularly, at least annually. Confidentiality: moderate. Event records could contain confidential science data and methods belonging to individual researchers. Integrity: high. The event archive represents an important record of LSST’s operational details. It could also be a target for deletion or corruption by an intruder wishing to hide tracks. Availability: low. Operations would not be threatened if access to this data archive were interrupted. 2.6.6.2 Network Routing Routing rules should be in place to enforce the Events communication topology Each LSST Subsystem, except for Visitor Networks, will have its own Event Broker, probably a main and a backup for failover. External communication (outside the local Subsystem): only allowed to other Event Brokers Internal communication (inside the local Subsystem): allowed with any host Draft LSST Security Plan 2.6.6.3 Risk Assessment 2.6.6.3.1 False Message Injection Page 33 of 49 False events could cause serious problems, especially if they propagate to sensitive Systems, or if there are large numbers of them. Extraneous logging False VOEvent alerts Erroneous status notifications ("telescope preparing for exposure", "nightly data transfer complete") False exceptions ("the camera is malfunctioning", "error during nightly processing", "DAC X has lost Y raw data") False events could originate from: Injection into a Broker by a rogue client A rogue Broker sending to clients or other Brokers A software bug or operational error in an otherwise properly functioning LSST subsystem 2.6.6.3.2 Interruption Event messages, like all LSST network traffic, are vulnerable to network disruptions. That subject is covered in network design and fault-tolerance design documents. 2.7 Operation Center The DM System Operation Center (DMSOC) will … see Ephibian document. 3 Sites 3.1 Summit and Base Site (JK - Ron Lambert review for accuracy) The Summit and Base Site (SBS) are situated on Cerro Pachon and La Serena, Chile, in facilities operated by the AURA. Page 34 of 49 LSST Security Plan Draft Figure 3.1 Summit and Base Sites They house: Summit o Observatory—observatory building, telescope, camera o The mountain end of the Observatory Control Subsystem (see section 2.1) Summit-Base Network Link o Connects Summit to Base site Base Site o The base end of the Observatory Control Subsystem (see section 2.1) o The Base Facility, comprised of: Archive Operations Subsystem (see Section 2.2)—responsible for a full archive of raw data Distributed Processing Subsystem (see Section 2.3)—responsible for real-time processing o The Chilean Data Access Center Community Service Subsystem (see Section 2.4)—responsible for Chilean public data access Visitor Network (see Section 2.5) The Summit Network provides the infrastructure used by systems such as the Observatory Control Subsystem to support the safe acquisition, transport, control and other operational activities related to LSST data and provides the communications path down to the La Serena Base Site, from whence the data is passed on to the other major LSST subsystems such as Data Access Centers and LSST data end users. The goal of LSST Summit/Base Site security is to provide physical and logical defense in depth to maintain the confidentiality, integrity and availability of LSST data and resources at the LSST Pachon and La Serena facilities, along with the long-haul communications infrastructure that connects LSST Pachon to the La Serena Base Site and beyond. This protection requires the prudent application of industry best practices and technologies in security as available at the moment of system acquisition and design. 3.1.1 SBS Responsibilities Most responsibilities are covered in Common Policies—Responsibilities, section 1.1 of this document. Draft 3.1.1.1 LSST Security Plan Page 35 of 49 Partner Institution: AURA The summit and base are housed in facilities operated by the AURA. AURA and LSST have signed a Cooperative Agreement that covers security responsibilities and staffing. 3.1.2 SBS Physical Operating Environment The Summit Site is located on Cerro Pachon in Chile, at 3300m altitude, roughly 40km into the AURA property shared between Cerro Tololo Interamerican Observatory, Gemini, SOAR and other tenants. Physical access to this site is restricted by the site's remoteness from populated areas as well as guards at the CTIO/AURA gate at the entrance to the AURA property from public access roads. The La Serena Base Site is located at the CTIO/AURA facilities in La Serena, Chile. Physical access to LSST resources at this site is controlled by guards and physical technical access controls into the LSST offices and computer center. 3.1.3 SBS Subsystems The LSST Summit and Base Site network infrastructure is defined in the LSST Network Design (document-7354). The network is safeguarded and controlled by various authentication solutions and monitored through network and host based Intrusion Detection Systems and centralized log file analyzers. Other than network devices that are specifically part of other LSST subsystems such as the Observatory Control Subsystem or the Camera System, the LSST Summit infrastructure also includes necessary system servers such as DNS, authentication servers, system log servers, intrusion detection sensors, and monitoring console systems. Further, the Summit network communications infrastructure provides a tunneled, enforced transmission path between acquired LSST data and the La Serena Base Site via leased communication lines, with the La Serena and Pachon networks falling within very similar Security Policy domains connected via a so called "Metropolitan Area Network". Within La Serena, there is a dedicated, enforced communication path for LSST data to the local Chilean DAC at La Serena. Confidentiality: moderate. While the confidentiality of the SBS should be protected, leakage of operational information or live data would not seriously impede LSST’s operations. Integrity: high. LSST’s highest security priority is the integrity of its scientific data; if that is called into question, it throws doubt on LSST’s central mission. Availability: moderate, with the Summit network requiring somewhat higher availability than the La Serena Base Site. A preliminary risk assessment of the expected elements of the Summit and La Serena LSST network suggests that these portions of the LSST infrastructure will require medium Confidentiality, high Integrity, and medium Availability, with the Summit network requiring somewhat higher availability than the La Serena Base Site. Page 36 of 49 LSST Security Plan Draft 3.1.4 SBS Management, Operational, and Technical Controls Operational controls are the most important element of security at the LSST Summit facility, with the goal of preventing any security incident that might disturb the data taking and data flow missions of the telescope. Detecting and correcting any anomalous events is also very important. The La Serena Network's mission is to support the Summit Network, and to provide the operational communication infrastructure necessary to safely send the LSST data on to Archive Sites and the LSST DACs. 3.1.4.1 Access Control The Summit network is the most private and protected network in LSST’s infrastructure. Only authorized LSST principals shall be granted access to it, and only from authorized hosts. As a purely internal subsystem, no external access will be granted. Remote user access will be limited to fixed gateways such as SSH servers or terminal servers, so that direct remote access can be avoided. Access to the Summit network is governed by Common Policies—Access Control, Section 1.4.2. In particular, access shall be granted only to specific LSST personnel. Individual LSST Applications software at the Archive and DACs will need to have network access to the LSST Summit and Base Site resources. These Applications shall have security designed in and shall use secure and enforced communications paths into the LSST Summit/Base Site facilities using security packages such as SSL or SSH tunnels. 3.1.4.2 Awareness and Training LSST Support personnel both at the Summit and Base Site shall complete the necessary security training to operate LSST facilities safely and to be aware of security concerns. They shall be required to sign off on LSST Security Policy documents including those addressing "acceptable use" and LSST Summit/Base Site monitoring policy, and that this acceptable use and monitoring policy shall be reinforced by logon banners when access to the LSST Summit/Base Site Network is made. 3.1.4.3 Audit, Accountability, and Monitoring There shall be centralized security logging and monitoring facilities for both the LSST Summit and Base Site Networks that will collect traffic, system log and other audit records from all critical computers and devices, including routers and firewalls. The centralized security logging and monitoring servers will provide data mining in order to analyze traffic and logs for intrusions and anomalies, and automatic alarm/event notification capabilities to alert personnel to the occurrence of significant events. Candidate technology would include event reporting features that are capable of near-real-time reporting and automatic alarms via local console, email, pagers and cell phones. Critical LSST Summit and Base Site computers shall be configured to periodically run integrity-checking tools and send their results to the central monitoring servers. Along with the host based intrusion described above, the LSST Summit and Base Site Network shall also support network based intrusion detection, using a subsystem which includes non-invasive network packet sensors that sample network traffic, and central monitoring servers that collect and analyze these packets to discover unusual network activity, security incidents or other events of interest. Draft 3.1.4.4 LSST Security Plan Page 37 of 49 Identification and Authentication Two-factor authenticated VPN and/or SSH access shall be required to access Summit/Base Site computers and network resources both from over the network and from within the LSST Summit or Base Site Network, through a centralized Identification Management system of a level To Be Determined, ranging from a simple authentication server to a more sophisticated single sign on (SSO) solution. The identification and authentication process shall be logged and correlated with other audit information including logs provided by physical technical access control devices necessary to gain access to La Serena server rooms. 3.1.4.5 Physical and Environmental Protection The LSST Summit and Base Site infrastructures will need additional protection in the layout of cable trays and equipment positioning and isolation due to the possibility of earthquakes at both sites. While tsunamis are possible in La Serena, the actual location of the CTIO/AURA facilities at 90m above sea level and 3.5km distant from the seashore should minimize the risk posed by this threat if historical records are any guide. The LSST Summit and Base Site facilities will need well sized UPS and power conditioning as historically in La Serena and at the AURA telescope sites power outages and brownouts are fairly common. Disaster recovery plans are documented in the LSST Operations Plan. 3.2 Archive Site The Archive Site (AS) will: House a full archival copy of LSST’s raw observational data Perform nightly processing and processing for annual data releases Provide services for the public and researchers House a co-located Data Access Center 3.2.1 AS Security Objectives The Archive Center’s security objectives differ for each of its subsystems; for more details, see DMS Subsystems, Section 2 of this document. 3.2.2 AS Responsibilities The Archive Center will be operated cooperatively by NCSA and LSST. 3.2.3 AS Physical Operating Environment The Archive Center will be housed in Champaign, Illinois, on the campus of the University of Illinois, at the Illinois Petascale Computing Facility, which will also house the Blue Waters supercomputing cluster (to be commissioned in 2010) and other NCSA-operated supercomputers. The facility will be operated by NCSA. Physical access to the building will be restricted, controlled by NCSA. Page 38 of 49 LSST Security Plan Draft 3.2.4 AS Subsystems The Archive Site will house: An Archive Operations Subsystem (AOS) A Distributed Processing Subsystem (DPS) A Community Service Subsystem (CSS) A Visitor Network Event Brokers as part of the Event Messaging Component (EMC) 3.3 Standalone Data Access Centers The LSST project will operate co-located DACs at the Archive Site and Base Site. Subject to expanded or external funding, one or more Standalone Data Access Centers (DACs) will be commissioned. Each Standalone DAC will house: A Community Service Subsystem (CSS) An Archive Operations Subsystem (AOS) A Visitor Network Event Brokers as part of the Event Messaging Component (EMC) Standalone DACs are similar in organization to the Archive Center, although the scale may differ; for example, a DAC’s AOS will be smaller, housing only a fraction of the archived LSST data, rather than a full copy, and caching frequently-accessed data retrieved from the main Archive Center. Draft LSST Security Plan Page 39 of 49 Figure 3.2 Network Perimeter for a Data Access Center 3.4 Headquarters and Education and Public Outreach Center The DM Headquarters Site will house an Education and Public Outreach (EPO) Center (see section 2.8) and a DM System Operations Center (DMSOC) (see section 2.7). Refer to EPO use cases/estimates. TBD. Contain a CSS only? Event Brokers? (JK – Will contain a DM System Operations Center (DMSOC, refer to Ephibian concept document. Will also feed/house EPO database (KB: user-writable by the external EPO community), refer to EPO use cases/estimates) Page 40 of 49 LSST Security Plan Draft 4 Threats Introduction Physical threats are covered in a separate document about operational risks. LSST cyber-threats include both unintentional acts (inadvertent data entry) and deliberate actions (network based attacks, malicious software upload, unauthorized access to confidential information). LSST is a high profile project providing data to the public. Therefore, there is little risk of public interest in obtaining LSST public data through illegal means. However, the nature of LSST’s computing resources (massive storage, processing, and network capability) will be an attractive target for a variety of reasons. Threat Source Public User Hacker, Cracker Motivation Unintentional Denial of Service (DOS) Challenge Ego; bragging rights Rebellion Threat Actions Organized computer criminals Monetary gain Hijacked resources Insiders (poorly trained, disgruntled, malicious, negligent, dishonest, or terminated employees, graduate students, scientists, etc.) Curiosity Ego Intelligence Monetary gain Revenge Unintentional errors and omissions (e.g., Denial of Service through automated queries or poorly constructed queries Unauthorized system access System intrusion Web page defacement Warez storehouse Social engineering Vector for further internal and external attacks (account subversion, DOS, etc.) Intrusion via automated virus, worms, etc. Unauthorized system access System intrusion Blackmail Data storehouse (warez, porn, etc. Malicious code (e.g., virus, logic bomb, Trojan horse) System sabotage Vector for further internal and external attacks (account subversion, DOS, etc.) Phishing Blackmail Browsing of proprietary information Computer abuse Unauthorized system access Fraud and theft Input of falsified, corrupted data Draft LSST Security Plan data entry error, programming error, denial of service) Page 41 of 49 Malicious code (e.g., virus, logic bomb, Trojan horse) System bugs System intrusion System sabotage Denial of Service through automated queries or poorly constructed queries 4.1 Vulnerabilities 4.1.1 Known Sources As with any organization implementing large-scale Information Technology resources, LSST faces human threats to infrastructure and availability. Human threat motivations differ but impacts vary primarily based on the type of access they have or can obtain. Human threats utilize a toolset based on their level of access, permissions, and technical capability. The following are common vulnerability classes posed by various human threats: Shoulder surfing Social engineering Brute force password guessing Data deletion/corruption Phishing Spam Unauthorized remote access Denial of Service Buffer overflow Cross-site scripting SQL injection Virus/worm/malware Unauthorized privilege elevation Application exploit OS exploit 4.1.2 Other Sources of Vulnerabilities The technical and nontechnical vulnerabilities associated with an IT system’s processing environment can be identified via the information-gathering techniques including: Questionnaire On-site Interviews Document Review Automated Scanning Tool LSST will conduct an annual review of other industry sources (e.g., vendor Web pages that identify system bugs and flaws) in preparing for the interviews and in developing effective questionnaires to identify vulnerabilities that may be applicable to specific IT systems (e.g., a specific version of a specific operating system). In addition, LSST designers and administrators will use resources available on the Internet as another source of information Page 42 of 49 LSST Security Plan Draft on known system vulnerabilities posted by vendors, along with hot fixes, service packs, patches, and other remedial measures that may be applied to eliminate or mitigate vulnerabilities. Documented vulnerability sources that will be utilized in an on-going vulnerability analysis include, but are not limited to, the following: The IT system’s audit reports, system anomaly reports, security review reports, and system test and evaluation reports Vulnerability lists, such as the NIST I-CAT vulnerability database (http://icat.nist.gov) Security advisories, such as FedCIRC and the Department of Energy’s Computer Incident Advisory Capability bulletins Vendor advisories Commercial computer incident/emergency response teams and post lists (e.g., SecurityFocus.com forum mailings) Information Assurance Vulnerability Alerts and bulletins for military systems System software security analyses. Note that the for LSST, the types of vulnerabilities that will exist, and the methodology needed to determine whether the vulnerabilities are present, will vary depending on the phase of development: During the R&D phase, the search for vulnerabilities should focus on the organization’s security policies, planned security procedures, and system requirement definitions, and the vendors’ or developers’ security product analyses (e.g., white papers). This is the current LSST phase. During the Construction phase, the identification of vulnerabilities should be expanded to include more specific information, such as the planned security features described in the security design documentation and the results of system certification test and evaluation. This is the next phase for LSST. During the Commissioning and Operations phases, the process of identifying vulnerabilities should include an analysis of the IT system security features and the security controls, technical and procedural, used to protect the system. This will be the final phase for LSST. 4.2 Control Analysis The goal of this step is to analyze the controls that are planned for implementation by LSST to minimize or eliminate the probability of a threat’s exercising a system vulnerability. The following sections discuss control methods, control categories, and the control analysis technique. 4.2.1 Control Methods Security controls consist of the use of technical and non-technical methods (e.g., firewalls as well as processes and procedures). Technical controls are described in section 1.1.12 but are summarized below. Process and procedural controls are not completely defined yet but will include Management Security and Operational Security. 4.2.2 Technical Security Controls Technical security controls are described in more detail in section 1.1.12 but will consist of: Draft LSST Security Plan Page 43 of 49 Access Control Lists Firewalls Network Intrusion Detection/Prevention Systems (NIDS/NIPS) Host-Based Intrusion Detection/Prevention Systems (HIDS/HIPS) Logical Network Separation Virtual Private Network (VPNs) Operating System Access Controls Application Access Controls Application Proxies Remote Gateway Systems Demilitarized Zones (DMZs) Encryption One-time Password Mechanisms Anti-virus/anti-malware software Anti-spam appliance/service 4.2.3 Management Security Controls Management security will consist of: Assignment of responsibilities Continuity of support Incident response capability Periodic review of security controls Personnel clearance and background investigations Risk assessment Security and technical training Separation of duties System authorization and reauthorization System or application security plan 4.2.4 Operational Security Controls Operational security will consist of: Control of air-borne contaminants (smoke, dust, chemicals) Controls to ensure the quality of the electrical power supply OS and application patch application Data media access and disposal External data distribution and labeling Facility protection (e.g., computer room, data center, office) Humidity control Temperature control Control of workstations, laptops, and stand-alone personal computers 4.3 Likelihood Determination The likelihood that a potential vulnerability could be exercised by a given threat-source can be described as high, medium, or low. These are described below. Page 44 of 49 LSST Security Plan Draft High: The threat-source is highly motivated and sufficiently capable, and controls to prevent the vulnerability from being exercised are ineffective. Medium: The threat-source is motivated and capable, but controls are in place that may impede successful exercise of the vulnerability. Low: The threat-source lacks motivation or capability, or controls are in place to prevent, or at least significantly impede, the vulnerability from being exercised. 4.3.1 Vulnerabilities from Human Threats. Breach or loss of data or computing resources from hacker/cracker/criminal computer organization – LOW. The LSST facilities and systems within the scope of this document will not have significant public exposure. There will be extremely limited port and protocol exposure and these will be heavily protected by multiple, powerful technical control mechanisms (e.g., VPN, one-time passwords, proxies, secure gateways, intrusion detection monitoring, etc.), plus procedural controls (i.e., timely patch application, etc.). Breach or loss of data or downtime due to insiders – LOW. LSST controls to mitigate insider threats include one-time passwords for administrative accounts, logical network separation, access control lists, and full data replication. Motivation for most insider threats is not considered high and the control mechanisms planned for implementation (both technical and non-technical) should mitigate the threats sufficient to warrant a low likelihood of impact. 4.4 Impact Analysis The adverse impact of a security event can be described in terms of loss or degradation of any, or a combination of any, of the following three security goals: integrity, availability, and confidentiality. The following list provides a brief description of each security goal and the consequence of its not being met. Loss of Confidentiality: System and data confidentiality refers to the protection of information from unauthorized disclosure. The impact of unauthorized disclosure of confidential information can range from the jeopardizing of national security to the disclosure of Privacy Act data. Unauthorized, unanticipated, or unintentional disclosure could result in loss of public confidence, embarrassment, or legal action against the organization. Loss of Integrity: System and data integrity refers to the requirement that information be protected from improper modification. Integrity is lost if unauthorized changes are made to the data or IT system by either intentional or accidental acts. If the loss of system or data integrity is not corrected, continued use of the contaminated system or corrupted data could result in inaccuracy, fraud, or erroneous decisions. Also, violation of integrity may be the first step in a successful attack against system availability or confidentiality. For all these reasons, loss of integrity reduces the assurance of an IT system. Loss of Availability: If a mission-critical IT system is unavailable to its end users, the organization’s mission may be affected. Loss of system functionality and operational effectiveness, for example, may result in loss of productive time, thus impeding the end users’ performance of their functions in supporting the organization’s mission. Draft LSST Security Plan Page 45 of 49 4.4.1 Magnitude of Impact Definitions High: Exercise of the vulnerability (1) may result in the highly costly loss of major tangible assets or resources; (2) may significantly violate, harm, or impede an organization’s mission, reputation, or interest; or (3) may result in human death or serious injury. Medium: Exercise of the vulnerability (1) may result in the costly loss of tangible assets or resources; (2) may violate, harm, or impede an organization’s mission, reputation, or interest; or (3) may result in human injury. Low: Exercise of the vulnerability (1) may result in the loss of some tangible assets or resources or (2) may noticeably affect an organization’s mission, reputation, or interest. 4.4.2 Magnitude of Impacts due to Vulnerabilities Loss of Integrity – LOW. The effect of a successful human threat breach to data integrity is anticipated to be low. Full daily data replication between multiple facilities and multiple integrity checks would ensure that any core data that becomes compromised could be discovered and corrected or restored unless it was compromised before any replication occurred. In that event, impact would be a potential loss of data for one day or however long the hacker/cracker/criminal remains undetected. Network control mechanisms and procedures are expected to be in place such that data flows from Summit to Base to local data store to main archive but with access restricted to pipeline data. Logical network controls should contain a breach to a facility. Most harmful would be a breach at a Chile facility since those facilities will contain the most recent data and a full copy of all data. Each day, new data will be replicated to the main archive center in the US. Even if all Chile data were corrupted or lost, the data at the US facility could be used to reconstitute all data except the data lost since the breach occurred. Loss of Availability – MEDIUM. The impact of a loss of availability is based on the facility where a breach occurred. If either full data store facility (Chile or US) were taken offline or became unavailable due to data loss, there would be an impact to those public users requiring access to archival data. Assuming a vulnerability was exploited to compromise the availability of a single facility, the remaining facility could provide sufficient availability, albeit with some degradation of performance, so that scientific and public availability would not be significantly affected. However, if all data at a facility were compromised and needed to be restored, the process of rebuilding complete systems from backup and then full repopulation of the complete data set across the available communications links would be highly time consuming; it would take more time the longer the LSST systems are in operation as the full data set will grow enormously over time. It could take a considerable amount of time for full data restoration especially with normal operational data flows occurring at the same time. The end result could be degraded performance for a lengthy period of time. Loss of Confidentiality – LOW. LSST data is generally not confidential. The primary confidentiality concerns are based on proprietary academic researchers’ data processing. A loss of confidentiality would likely lead to a loss of confidence in the administration of the system, embarrassment, and loss of reputation to LSST among researchers. While these Page 46 of 49 LSST Security Plan Draft losses could be considerable to LSST prestige and to researcher peace of mind, the general public would not be directly affected and amateur astronomers could continue personal research since the integrity of core LSST data would not be impugned. 4.4.3 Mitigation Strategies 4.4.4 Vulnerabilities from Human Threats The LSST facilities at the summit, base, and national archive center are subject to vulnerabilities due to human threats from hackers/crackers, computer criminal organizations, and insiders. Vulnerability: Shoulder surfing Mitigations: Security awareness training Physical security One-time password mechanisms for administrative functions Residual Risk: LOW Vulnerability: Physical theft of resources or media Mitigations: Physical security Procedures require visitor escort Procedures limit visitor access Logically separated visitor network segment Residual Risk: LOW Vulnerability: Social engineering Mitigations: Security Awareness training One-time passwords Residual Risk: LOW Vulnerability: Brute force password guessing Mitigations: One-time passwords OS access controls Firewalls and network ACLs (for remote attempts to obtain a password file) Residual Risk: LOW Draft LSST Security Plan Vulnerability: Data deletion/corruption Mitigations: Operations training (against accidental deletion/corruptions) Data replication and restoration processes Incident response process Residual Risk: LOW Vulnerability: Phishing Mitigations: Security Awareness training Anti-virus/anti-malware software Residual Risk: LOW Vulnerability: Spam Mitigations: Mail Transfer Agents configured not to allow open relays Anti-spam appliance/service Residual Risk: LOW Vulnerability: Unauthorized remote access Mitigations: Firewall limited exposure of remote access ports and protocols Encrypted VPN connectivity Strong authentication (one-time password) Network Intrusion Prevention systems Application access controls Remote gateway desktops Residual Risk: LOW Vulnerability: Denial of Service Mitigations: Firewall limited exposure of remote access ports and protocols Network Intrusion Prevention systems Host configurations to limit syn flood connections Incident Response process Residual Risk: LOW Vulnerability: Buffer overflow Mitigations: Page 47 of 49 Page 48 of 49 LSST Security Plan Applications written to limit and verify all user input fields Application proxy devices Short turn-around patch application Network Intrusion Prevention systems Residual Risk: LOW Vulnerability: Cross-site scripting Mitigations: Applications written to limit and verify all user input fields Applications written with cookie security Residual Risk: LOW Vulnerability: SQL injection Mitigations: Applications written to limit and verify all user input fields Applications written with strong typing whenever possible User input translated to parameterized statements or escaped at minimum Residual Risk: LOW Vulnerability: Virus/worm/malware Mitigations: Anti-virus/anti-malware software required on all LSST owned Windows-based workstations and servers Logical network segmentation to limit propagation Network access controls lists to limit propagation Network Intrusion Prevention systems Host-based Intrusion Prevention systems Residual Risk: LOW Vulnerability: Unauthorized privilege elevation Mitigations: Firewalls (for protection from external threats) Network Intrusion Prevention Systems Host-based Intrusion Prevention Systems OS access controls Logical network separation Network access control lists Application access controls Separation of duties Residual Risk: LOW Draft Draft LSST Security Plan Page 49 of 49 Vulnerability: Application exploit Mitigations: Applications written to limit and verify all user input fields Applications written with cookie security Applications written with strong typing whenever possible User input translated to parameterized statements or escaped at minimum Short-turnaround application patch application from patch release Residual Risk: LOW Vulnerability: General OS exploit Mitigations: Minimal OS features necessary enabled Short-turnaround application patch application from patch release Residual Risk: LOW The controls provide mitigation for the vulnerabilities associated with human threats and provide a high degree of protection against data loss, corruption, and confidentiality however there is always residual risk associated with human threats, especially insider threats and remote attacks based on zero day exploits.