Project S.I.N. A comparison of network monitoring systems Brent Grover Edward Schilla Lucas Schill Michigan Technological University Abstract— In comparison of legacy Nagios and recent forks Icinga and Shinken, the question of enterprise solution efficiency comes into play. The project focuses on performance and usability of these three network monitoring suites on enterprise simulated computing environments. The goal is to create a recommendation to firms who are looking into a solution to suit their network monitoring needs. Keywords—component; formatting; style; styling; insert (key words) I. INTRODUCTION The goal of project S.I.N. is to compare the three infrastructure monitoring suites in respect to the demands of a corporate environment. To accomplish this, we will create a network that is designed to simulate the complexities and requirements of a full scale production environment. Once completed we will deploy and test the three suites via a rubric that we will develop. This test will measure many of the features and qualities of each, from easeof-use to efficiency. II. ENVIRONMENT OVERVIEW A. Current Setup As stated, NFS, PXE, and DNS are managed on one of two IBM servers while pfSense is managed on the other. Upgrading or changes to this will be the last thing the project seeks. Three Dell 2650 1U servers installed with RHEL 6.3 and dedicated one each to Nagios, Shinken, and Icinga. Two of them are currently installed and running at the time of this writing. Upgrades to these servers is of utmost priority, as they are the root of all testing metrics. Previously mentioned Netgear and Cisco networking hardware is of least importance on the upgrade list, as it will be difficult to justify going above 100mbps interfaces in today’s local enterprise networking environments, except for possible interface upgrades to uplinks. The primary site consists of a pfSense firewall, with one NIC connected to MTU’s network, one NIC connected to the internal testing networking, and a final nice connected to a client network. The internet network consists of a Netgear FSM7352S and a CISCO 3550 48 port. These two switches are both 100Base -T. The client network is made up of a Netgear FS105 5 port switch and a Meraki MR12 wireless access point. B. Goal Our planned network consists of two sites, the primary site that holds most of our equipment and our client machines, and a remote site that will be used to test specific features that need to be present to successfully monitor an enterprise system. The primary site is located at Michigan Technological University, behind a pfSense box that serves as DHCP and NAT for our network and uses the IP provided to us by MTU for internet access. The client site of this network is located on a separate interface and subnet to isolate the test environment, it contains two desktop machines and an access point for other devices. The other part of this network contains most of our testing equipment including our three primary SIN nodes that the software will run on and many clients. The remote site will contain one SIN node and numerous clients. The location for this site is still undetermined. This location’s main purpose is feature testing, with secondary focus. III. TESTING To properly test specific features such as clustering while assuring consistent resource access between the three soft wares we are going to run our SIN nodes as KVM hypervisors and divide up the available resources evenly between the KVM guest. Each node will have 3 virtual machines running on it, one for each software. This design also gives us the ability to save machine states and restore them during testing to keep the environment clean, and it is following the virtualization trend. OS and Software Our SIN nodes will be using Debian 6.0.6 (RHEL) 64bit and will have KVM software installed. The guests will use Debian 6.0.6 64bit but their software will vary depending on the requirements of Shinken, Nagios and Icinga. The client nodes will run numerous operating systems, including CentOS, Debian and Windows to fully test the software. Performance, feature testing through custom scripts. We will write scripts and run them and test features and etc. We will grade according to a fact based rubric and an opinion based rubric. C. pfSense The pfSense firewall runs on an IBM xSeries 340 and works as a NAT, DHCP, Firewall, and DNS relay. Each of the two internal interfaces has a DHCP server running on it to serve out dynamic IPs and PXE information when needed, although all permanent equipment uses static assignments. A Virtual private network (VPN) is configured to allow access from outside the network, and a NAT Forward for Secure Shell (SSH) is setup to provide additional access to internal resources. D. PXE/WDS With the large amount of clients and servers on the project’s network, it was decided that a network based installation solution would be the most efficient way to add operating system images to the client and server computers. The Linux based computers receives the initial Preboot Execution Environment (PXE) image from the PXE server which loads the base operating system and then loads the configuration file. This configuration file streamlines the installation process by providing the options such as Network File System (NFS) server location, usernames, passwords, and additional packages to install. The remaining installation files for the RHEL installations are retrieved from the NFS server by the client. For the Debian installations, additional files are retrieved from the Linux User Group’s servers at Michigan Technological University. To make diagnosing clients and servers easier, network based tools such as memtest and gparted were also added to the PXE server. Clients utilizing the Windows 7 operating system receive their images from the Windows Deployment Server (WDS). A Windows 7 image was placed on a client machine, fully updated to include service pack one, and captured via the WDS. This updated image ensures that minimal updating is required when the new clients are imaged. To streamline the installation process for Windows 7 clients, the Microsoft Deployment Toolkit (MDT) along with Automated Installation Kit (AIK) were installed on the WDS. MDT and AIK allow for the creation of task sequences so that various scripts can be run during the installation. This also creates a method for zerotouch deployment, all that needs to be done is start the PXE session on the client and it will finish the imaging process automatically. To run both the Linux PXE server and the WDS server on the same network, some additional configuration was needed. Due to the fact that WDS needs to be the primary network boot server on the network to properly serve out its installation files, PXELINUX was added to it and set to override the default PXE service. A new PXE menu was then created on the WDS to have options for both the WDS and a pointer to the Linux PXE server. E. Clients Client PCs are installed with the latest debian stable net install, 6.3, and include the packages sshd as well as vim on top of just the base image. Static IP addresses are configured to ensure control of testing. A formal inventory of these machines is kept in a spreadsheet including the processor type, speed, memory size, and storage size. This information is important for both capability of the machines but also to test against scripts that will be run in the testing phases. Currently, there are ten dell desktop computers with Intel Pentium 4 processors ranging from 2.26 Gigahertz (GHz) to 3.80GHz and all include between 512 Megabyte (MB) and 1 Gigabyte (GB) of DDR memory. The 11th machine is an AMD Athlon xp pro, which is very similar in performance. All of the clients also have integrated network interface controllers. F. Servers Three Dell 2650 2U rack mount servers are the brains of the operation. Each is dedicated to its own flavor of network monitoring system, and includes a fresh, updated image of Red Hat Enterprise Linux (RHEL). RHEL is some of the most widely used Linux server operating system software in the computing industry and is therefore the most logical choice for this project. Most integration of the network monitoring systems tested has specific support in the RHEL community and the systems are also optimized for such enterprise software. V. Miscellaneous Project Information G. A. Project Goals Motivation behind Project S.I.N. comes from issues with Nagios being claimed to not have features or lack of efficiency with feature that Icinga or Shinken may have advantages in. Testing of network utilization, server utilization, client load, ease of management, and other factors will help firms make informed decisions about integrating these tools into their systems. H. B. Setbacks First and foremost has been the inexistence of support from Nagios, Shinken, and Icinga. Shinken, as of now, is only community based. While this is positive for correctness, response time for issues may be a bit slower than expected. The next largest inconvenience comes with the acquisition of operating system images, as this has taken around three weeks worth of time away from the project. Thirdly, more hardware is needed, whether it is new or not, for client use. As stated in the “Goals” section of “Environment”, creating multiple virtual machines as clients would be incredibly beneficial, but at this point even expanding on our physical machines running Pentium 4 processors would improve the testing environment. The last concern that will eventually become a setback during the testing phase is the age of the Dell 2650 servers. Server age will depreciate the enterprise environment standard that has been set and will not allow for accurate testing measurements. I. D. Website The group’s website was designed to be the frontend to possible sponsors and others interested in the project. A project plan and network diagram are available in PDF form for easy access. Weekly updates , contact information, and the budget are also available on the website. It was designed to be sleek and clean so that it is easy to read. The Twitter Bootstrap framework was utilized to layout and style the website.