MPRIME Preliminary Architecture

advertisement
MPRIME Preliminary Architecture
July 11, 2002 (Draft #3)
Background
The Michigan Program for Research Information Management and Education (MPRIME) is a project, which aims
to improve the protection of human subjects in research by improving the administration and approval process
for projects and protocols involving human subjects.
In April 2003, MPRIME issued an RFP for products which support core processes that impact human subject
research compliance. In July 2003, Webridge Extranet Portal emerged as the leading product in the RFP
selection process.
This document provides a high-level overview of the technical architecture that would be needed to support a
Webridge deployment. Details of the architecture are likely to change during implementation, but this
document represents a likely scenario of how the architecture will be deployed.
Webridge architecture summary
More than any product that we have seen to date, the Webridge software architecture is based completely on
Microsoft standards. Company representatives proudly state that they have strategically aligned their
technology with Microsoft, and believe this alignment has proven to be wise and has contributed to their
corporate success.
The Webridge product is primarily web-based, and is based on a classic n-tier architecture. The components of
the architecture include:

The Browser
All end-users, administrators and developers will use a web browser to interact with the Webridge
applications. Like most web applications, the browser primarily handles presentation of the application. No
business logic occurs on the client.
Some Webridge development cannot be done through a browser, and must be completed using Webridge
“fat client” tools. U-M anticipates that this type of development is rare, and that the vast majority of
development will be done through a web-interface.

Web Server
Microsoft Internet Information Server (IIS) is used for all web communication. IIS’ primary role is to deliver
and receive HTTP/HTTPS traffic to and from the client. Very little Webridge processing occurs within IIS.
The exception is an Active Server Page (ASP) that is used to route requests properly to the application
server.

Application Server
The Webridge application server is the heart of the Webridge system. The application server is responsible
for nearly all Webridge business logic. Responsibilities of the application server include:

Retrieving business logic from the database, and then executing all business logic.

Sending all query and update requests to the database.

Construction and deconstruction of objects in memory through the use of XML and database calls.

Developing HTML for presentation of to end-user

Management of memory and disk cache.
The application server is build with Microsoft Transaction Server. Originally written in Java, the Webridge
application server is now been ported to “J#” (“J sharp), Microsoft’s modified version of Java. The
Page 1 of 16
application leverages many of the core services that are available on the Microsoft Windows Operating
System.

Database Server
Webridge relies on Microsoft SQLServer for all database services. Not only is traditional application data
stored in the database, but also much of the application business logic is also stored in the database. This
is similar to the way PeopleSoft stores the business logic within “PeopleTools” tables.
All business logic is stored within the database in XML format. The application in turn can then retrieve the
XML, and construct objects as necessary.
Webridge also takes advantage of SQLServer stored procedures. Stored procedures process some business
logic, and are used to support other interfaces within Webridge, most notably the interface between the
Corpus and the database. The stored procedures can be modified by the U-M, but Webridge technical
representatives discouraged this.

File Server
A significant amount of the application content in Webridge is stored within traditional document files (e.g.
Microsoft Word, Adobe Acrobat). Document files can be stored on any Microsoft file system. Optionally, a
file server can be added to the environment to handle file storage and retrieval.
Webridge refers to the repository of documents as the “Corpus.” The Corpus is a simple set of directories
for storing files, which can be navigated by the Webridge application server to retrieve files.

Index Server
Webridge applications allow end-users to do search structured (database tables) and non-structured
(documents) data. Non-structured searches are handled through the use of the Microsoft Index Server.
The MS Index Server maintains a file-based index of all documents within the Corpus. This allows Webridge
to support complex searches by end-users.
Customization and configuration
Webridge is an extensible environment. Webridge customers can use Webridge-provided tools to modify
various components of Webridge applications. Webridge classifies modifications in two ways:

Configurations
Through the use of Webridge web-based tools, developers can modify aspects of the Webridge application,
such as web forms and work flows. Webridge uses the term “configuration” to describe these
modifications, because these types of changes are non-programmatic in nature. The implication of the term
“configuration” is that the modification is not very significant to the application. Unfortunately, this is
misleading because some modifications can have significant impact. For example, some configurations can
result in changes to the underlying database structures, including adding new data types. Configurations
are “protected” during upgrades to the Webridge application framework. Extensions to the database
structure are rolled over untouched to the next upgrade. The web-based tools does allow scripting.
Scripting can be used to verify data, call external code, etc. Scripts are not “protected” during upgrades.
Scripts would need to be thoroughly verified during upgrades.

Customizations
Through the use of Webridge “fat client” tools, Site Designer and Entity Manager, developers can modify
various definitions that are stored within the database as XML definitions, and stored procedures.
Examples of common types of customization include the integration with outside systems and the definition
of complex nested reports.. Also if the SQL generated by Webridge is inefficient, it is possible for U-M staff
to add “hints” inside of stored procedures. The tools allow programmers to modify any data type, including
the “base data types” in the system. Customizations to base data types are not protected during upgrades
of the Webridge product and need to be tightly managed if they are undertaken. Ideally, U-M would not
modify these base data types, but define related data types to store required information.
Page 2 of 16
MPRIME databases requirements
An MPRIME implementation of Webridge would likely involve the following environments, each with their own
SQLServer database:

Production
This would be the primary Online Transaction Processing (OLTP) database for MPRIME. Mission critical
University data would be stored in this environment.

Reporting
A read-only database created by a Webridge-provided batch conversion program. While this environment is
“data warehouse-like,” it should not be considered the data warehouse. It can be better described as a mix
between a data warehouse and an operational data store (ODS). It’s not quite as well planned as the U-M’s
data warehouse, but it’s not an exact replica of production like the ODS.
The reporting data only contains structured data in the database. The Corpus (files) is not copied to the
reporting database.

Quality Assurance (QA)
A staging area, where customizations and configurations are tested before they are moved to the
Production and Reporting environments. At this point, the presumption is that this will be a productionsized environment which is replicated from the Production database a few times each year.

Development (multiple)
An environment used by a developer to customize or configure the Webridge applications. Each developer
has his/her own development environment. Development environments are stored locally on the
developer’s workstation.
It is important to note that the preceding list does not include many types of databases, which are in use, or
have been requested for other systems. For example: demonstration, stress testing, training, upgrade and user
“playground” will not be provided as part of MPRIME.
End-user desktops
Webridge supports both Internet Explorer (4 and greater) and Netscape (4.78 and greater). According to the
vendor, the product is based on HTML standards and therefore should not pose problems on different browsers
and operating systems.
The vendor did indicate one area where cross-platform compatibility is not available: users that wish to transfer
reports directly into Microsoft Excel can only do this on Internet Explorer and Windows computers. This
functionality will not work with Netscape or the Macintosh operating system.
Developer desktops
Browser-based development (configuration) can occur on the same platforms supported for end-users.
Fat-client-based development (customization) can only occur on Windows computers with the complete set of
Microsoft software (IIS, MTS, SQLServer).
It is also possible to provide the fat-client tools from Citrix. During implementation we may decide this is a
more efficient and effective model for distributing these tools.
Page 3 of 16
Architecture design points
In the following pages, the hardware and network architecture for the Webridge system is detailed. The design
of this architecture is based upon a number of fundamental decisions and assumptions. These include:
1. The ITS storage area network (SAN) should be used for all database and file storage (corpus).
Availability, performance, recovery and replication requirements are likely to require the robust SAN
architecture.
2. Any ITS server, which is connected to the SAN, should be in the ITS Secure Zone.
To maintain the security around the SAN network, access to the SAN is limited only to machines that are in
the ITS Secure Zone. This is a ITS standard that could only be changed through discussion with the ITS
Information Security Officer and the ITS Associate Vice President.
3. The web and application server need to reside on the same physical machine.
Webridge technical staff indicated web and application servers could reside on different servers, but that
they only support configurations where they are on the same server.
4. Production and Reporting databases will reside on the same server.
Reporting database will only used by a small set of people at a time, and probably does not warrant its own
complete set of hardware.
(Note: The index server and index files need to be on the IIS/Application
server.)
5. All web servers should reside in the ITS DMZ.
This is also a ITS standard. Since the purpose of web servers is to interact with users outside of the
firewall, the DMZ is the appropriate location for web servers.
Note, the combination of #1, #2, #3 and #4 above means that every MPRIME environment (e.g.
Production, Quality Assurance) will require two servers: a web/application server in the DMZ, and
file/database server in the Secure Zone.
6. Production and Reporting database instances can share one web/application server.
Webridge technical representatives indicated that it is common to see multiple databases connected to one
web/application server.
The benefits of sharing a web/application server include: fewer migrations, fewer upgrades, and fewer
physical boxes to administer.
The disadvantages of sharing a web/application server include: system competing for processing resources
and the systems are tied together for all upgrades and maintenance.
7. Redundancy of servers will not be pursued at this time.
For example, there is only one web server, one application server, and one database server. We will
purchase a hot spare that we can use in the event of a massive hardware failure. The hot spare server will
also be used as an upgrade server to test new Webridge releases without impacting the running Production,
QA, and Development environments.
8. No load balancing is needed, but SSL acceleration should be used.
Even though there will be only one web server, ITS’ will use its load balancing infrastructure to provide SSL
acceleration. This will offload processing from the MPRIME web server and help with performance. Other
than SSL acceleration, the load balancing infrastructure will not be leveraged.
9. No training environment will be necessary. Will use test for training purposes, limited hands on training
required for application.
10. No stress-testing and application tuning environment will be necessary. (Judy to confirm.)
Since little application tuning can be done inside of MPRIME, a separate “stress” database will not be
needed. Tuners will have to use the Quality Assurance environment to tune applications. Note, this could
Page 4 of 16
result in slower development schedules because developers and tuners will need to share common objects,
and therefore may not be able to work as efficiently.
11. Cosign will be used for authentication.
Technical representatives indicated that authentication exits are available which will allow the U-M to
implement Cosign. U-M has already developed the necessary Cosign code to work with Microsoft IIS
applications, so we believe integration of Cosign with Webridge should be very straight-forward. (What
about NDS for Medical School.)
12. No Citrix or Window Terminal Services will be needed.
All client and developer tools will be implemented through browsers and fat client installations. (Or should
the two fat clients be implemented on Window Terminal Services?)
13. MPRIME servers will be placed within existing M-Pathways production windows domain.
Sharing the M-Pathways domain will be more efficient, because more servers won’t have to be purchased
and administered to support the domain. Putting the Development/Quality Assurance servers in this
domain makes sense because some developers will not be ITS staff, so the ITS domain wouldn’t make as
much sense.
14. No central printing or online viewing will be needed.
All reports will be delivered through the Webridge interface, therefore no interfaces to Dazel and Vista Plus
will be needed.
15. No Electronic Data Interchange (EDI) will be needed.
There are no EDI interfaces with the Webridge system.
16. All interfaces to MPRIME are file-based.
All interfaces between MPRIME and other systems will be based upon the exchange of files. For example,
data needed from the PeopleSoft systems will be exchanged by files through standard mechanisms (SSH).
(Make this assumption. We will not know this for sure until we get into detailed design, but using our
current model , we believe this to be true)
17. No Virtual Private Networks (VPN) will be needed for interfaces.
Since all interfaces are within the U-M, no VPN will be needed to support secure transmission of interface
files.
18. Virtual Private Networks (VPN) may be needed to support developers outside of the ITS network.
It may be necessary for developers to share common file systems or other services. Also, some developers
need to have direct DCOM communication to the application server. In order to assure these transmissions
are secure, VPNs may need to be deployed.
19. Customizations will be managed through Microsoft Visual SourceSafe (VSS) software.
Developers will check-in, check-out and manage all objects through VSS. VSS requires file server access to
developer desktops.
20. Configurations will be managed through a manual process.
No tools exist to manage migrations of configurations. Webridge provides a method to export and import
many types of changes from one environment to another, but this process is not automated. The tools
allow for a “package” to be created, but then the packages are applied to each environment manually.
MPRIME staff will need to have a formal procedure for migrating these changes.
Note, some development objects cannot be migrated and need to be reapplied by hand in each
environment. (Cindy will research.)
21. No development will occur on the production server.
Page 5 of 16
Webridge technical staff described methods of using the production server for development. It involved
saving changes in a holding area before applying them to production. This approach will not be used,
because it will make synchronization with Quality Assurance and Development too difficult.
Page 6 of 16
Preliminary Hardware Plan
The following diagrams depict the hardware that will be used by MPRIME, and the software components which
will reside on each piece of hardware. Black circles (with numbers) indicate new servers which need to be
purchased to support MPRIME implementation. All other equipment is already in production within ITS.
Diagram #1: MPRIME Production
Production
M-Pathways
Backup Domain
Controller
Production
M-Pathways
Primary Domain
Controller
Microsoft Internet
Explorer
Microsoft Office
Applications
1
Production
Web/ Application Server
User
Webridge
"Pass Through"
Web Transaction
(ASP)
https
Web Server
(MS IIS)
https
Webridge
Application Server
(MTS)
Production
DB/File Server
Oledb
Production
(SQLServer)
UMich Cosign
ISAPI
Extention
MS Index
Server
Index of Corpus
Content
Batch Program
To Create DB
Stored Procs
oledb
Webridge
ISAPI
Extension
2
SMB
/Net
BIO
SP
orts
135
Reporting
(SQLServer)
-1 3 9
Production
Webridge Corpus
tcp
Production
ITCS Server
UMich Cosign
Server
The production MPRIME environment will be made up two servers: production web/application Server and the
production database/file Server. Not only does the database/file server house the production database, it also
houses the reporting database as well.
The MPRIME servers need to operate within a standard Windows domain. Instead of creating a new MPRIME
domain with the overhead of additional equipment and management, MPRIME will be placed within the existing
M-Pathways production domain. The “domain controller” servers near the top of the diagram represent some of
the key existing equipment from the M-Pathways domain that will be used by MPRIME.
During the second half of 2003, ITCS will place a Cosign server within the ITS machine room. MPRIME will use
this server as its primary authentication server.
Note, this diagram does not include all servers that will be used (e.g. Kerberos, DNS, Unicenter, etc.), but
instead includes only the primary servers that will make up the MPRIME environment.
Page 7 of 16
Diagram #2: MPRIME Development & Quality Assurance
Production
M-Pathways
Primary Domain
Controller
Microsoft Internet
Explorer
Webridge
"Pass Through"
Web Transaction
(ASP)
https
Microsoft Office
Applications
Web Server
(MS IIS)
Webridge
ISAPI
Extension
https
Webridge
Application Server
(MTS)
Webridge Site
Designer
oledb
4
QA DB
(SQLServer)
Dev DB
Stored (SQLServer)
Procs
Stored Procs
Index
Server
SMB/Net
BIOS P
orts
Index of Dev
Corpus Content
Index of QA
Corpus Content
Specialized
Developer
Desktop
Development & QA
DB Server
3
Development & QA
Web/ Application Server
Standard
Developer / Admin
Desktop
Production
M-Pathways
Backup Domain
Controller
135-139
Dev Corpus
(MS File Server)
QA Corpus
(MS File Server)
UMich Cosign
ISAPI
Extention
tcp
dcom
Production
ITCS Server
Webridge Entity
Manager
dcom
UMich Cosign
Server
The development and quality assurance environment is identical to the production environment but with these
relatively small changes:

The two MPRIME servers will be less powerful (probably only half of the CPUs of the production servers).

Some developers will connect to the application server using Webridge fat-client development tools over
dcom.

Separate corpuses must exist for development and quality assurance. (Cindy, we need to confirm that
this really works. I believe that this does work.)
Page 8 of 16
Diagram #3: MPRIME Network
ht
tp
MAIS
Machine Room
Firewall
ps
U-M
Network
s
DMZ
Network Switch
Specialized
Developer
Desktop
User Desktop
Standard
Developer/Admin
Desktop
Specialized
Developer
Desktop
dc
om
on
Standard
Developer/Admin
Desktop
htt
aV
PN
User Desktop
https
Internet
MAIS
Staff Network
Firewall
Secure Zone
Network Switch
ITCS Production
Cosign Server
Production
Database/File
Server
Load Balancing
(SSL Acceleration)
Test
Database/File
Server
Production
Web/Application
Server
https
om
dc
MAIS Staff
Network
MAIS
Storage Area
Network
Test
Web/Application
Server
Demonstration
Web/App/DB/File
Server
The network design for MPRIME is similar to other applications that are currently run by ITS with a few
exceptions. These exceptions include:

Fat client software will be deployed to some non-ITS desktops. This client software will require file server
access and communicate using dcom. The use of a VPN to secure the connection may be necessary.

Load balancing switch will only be used for SSL acceleration not for load balancing across multiple web
servers.

There will be some users of the systems that are not on the U-M network.
Page 9 of 16
Version control architecture
In order to meet its business requirements, the U-M will customize and configure various portions of the
Webridge applications. The following diagrams illustrate how different types of objects will need to get
migrated through each MPRIME environment.
An important assumption in these processes is that each developer has his/her own local development
environment. Multiple development databases create logistical problems, which require thorough management
of the customization process.
Diagram #4: Migration Path for Site Designer and Entity Manager Customizations
Production
Environment
SD & EM
Customizztions
(Zipped File)
QA
Environment
SourceSafe
Export
Webridge
Site Designer
Development
Environment
SourceSafe
Repository
Webridge
Entity Manager
Version control for Webridge customizations is handled through Microsoft Visual SourceSafe (VSS). This is a
full-featured version control product with many advanced features. The following steps describe a likely process
that would be used to develop a customization with Site Designer and Entity Manager and move them to
production.
1. Developer “checks out” desired objects from the VSS repository. The objects are locked, so other
developers cannot modify them until they are checked back in.
2. Developer uses Site Designer and Entity Manager to build and test customizations.
3. Developer moves customizations back into VSS repository.
4. Developer (or Migrator) exports changes to a “zip” file.
5. Developer (or Migrator) uses Webridge tools to apply the “zip” file to the Test server.
6. Testing is done and process is repeated until code is ready for production.
7. Developer (or Migrator) applies the “zip” file (or files) to the production environment.
8. Developer (or Migrator) unlocks the objects for use by other developers.
Page 10 of 16
Diagram #5: MPRIME Configuration Migration Path
Webridge
Browser-based
Configuration Tools
Production
Database
Webridge
Import/Export Utility
Webridge
Browser-based
Configuration Tools
QA
Database
Webridge
Browser-based
Configuration Tools
Development
Database
SourceSafe
Repository
The Webridge product does have some tools to allow export and import of changes, but there is no advanced
version control capability available. We plan to leverage VSS for this purpose and use it to be the keeper of the
official copy of the application. The following process describes the steps that would likely be used to migrate
configurations.
1. Checkout the project type definition from version control. If not using version control, the it is assumed that
the production store is the keeper of the official copy and you should export the project type from the
production server.
2.
Import the project type definition into a development store
3.
Make the required changed to the workflow via the web based administration forms and perform the
initial testing on that store.
4.
When satisfied that the changes are complete and accurate, export the project type definition from the
development store.
5.
Import the new project type definition into a Staging/Test server which is created from a recent backup
of the production store.
6.
Test the new version of the Project Type. If issues are discovered, return to step 3.
7.
When satisfied that the changes behave as intended, backup the production store then take the same
project type definition export that was imported onto staging and import it into production. It
is recommended that this is done during a period of inactivity on the production server as, depending upon
the nature of the changes, there may be a requirement to migrate data structures.
8.
Check in the project type definition export into version control so that it can serve as the master for the
next round of changes.
Page 11 of 16
Diagram #6: MPRIME Stored Procedure Migration Path
Production
Database
QA
Database
Script which
modified Stored
Procudure
Development
Database
If stored procedure changes are related to the Site Design and Entity Manager, then migration process is the
same as in Diagram #4. If the stored procedure changes are manual, such as tuning, then the following
process would most likely be used:
1. Tuner builds script to make changes in the development environment.
2. Tuner (or Migrator) applies script against Quality Assurance environment.
3. Tuner (or Migrator) applies script against Production environment.
These changes need to be carefully tracked, because any other changes to the stored procedures will write over
these modifications, so they will need to be reapplied.
Page 12 of 16
Implementation Challenges
In developing this preliminary architecture document, several challenges were identified which need to be
addressed during implementation.
Webridge specific challenges

New version control system
ITS systems are currently based on version control using Quest Stat software. MPRIME will need to rely
on a completely different version control software (VSS) and migration process. Relying on multiple
products and processes could result in some inefficiency and overhead in the organization.

Vulnerability of IIS, may adversely affect service level goals
Microsoft IIS has proven to be the single biggest target for security attacks over the past few years.
ITS has specifically avoided using IIS for enterprise systems for this reason. When IIS had to be used,
ITS has set firewall policies to limit access to U-M sites, which prevented attacks that originated from
other Internet sites.
If security attacks continue to be an ongoing problem with IIS, ITS may find it difficult to assure the
high service levels that are needed or desired for MPRIME.

Poor scaling options
Most applications today allow for the use of multiple web and application servers. Organizations can
add as many servers as necessary to meet performance targets. Webridge only supports a single
web/application server. Other than buying the biggest server possible, there is no way for hardware to
be acquired to help address performance problems.

Future of Macintosh support questionable
Webridge is firmly and closely aligned with Microsoft technologies. While it appears the current version
of Webridge can work with the Macintosh, there will always be a risk that future versions will diverge
from the Macintosh when and if Microsoft architectures become less standards-based.
Before proceeding with the purchase of Webridge, the U-M should recognize the distinct possibility that
other operating systems (e.g. Macintosh, Linux) may not work with future versions of Webridge.

No significant upgrade support tools
There are no upgrade support tools to assist an organization assess the impact of a new version on
their modifications. It will probably be necessary to have an accurate and thorough record of all
customizations to help with the analysis during upgrade planning. The magnitude of this challenge
depends on the scope of changes made by U-M (more changes, more risk), and if Webridge upgrades
prove to be implemented as Webridge staff have indicated (if they misrepresented the situation, more
risk).

Little control over generated SQL
The Webridge application server constructs almost all SQL on the fly, or from stored procedures. There
is little opportunity for application tuning. If Webridge generates inefficient SQL, U-M’s only course may
prove to be to inform Webridge and hope that they provide a patch.

Solution for home development needed
Since production support often requires developers to work from home or remote sites, it may be
necessary to employ Citrix or other technologies to provide access to tools, file systems and other
resources. (Note, at the present time we are assuming that no Citrix access is needed.)
Page 13 of 16
MPRIME specific challenges

Distributed development
Current MPRIME plans call for developers to exist in multiple University units (at least ITS, DRDA and
Medical School). Distributed development has proven difficult in the past and was abandoned on the
M-Pathways project. Well-defined processes, which assure adequate controls, will be necessary and are
likely to conflict with the approaches used in some University units.

File formats
Submission and review of document files is a key component of the MPRIME system. Sharing
documents between different units introduces software compatibility issues. For example, a document
may be stored in a version of Word or Acrobat which is available in one unit, but not available in
another unit. The problem becomes exaggerated over time, as files stored in the system become many
years old, there may not be software available to open them.

No data growth or sizing metrics available
No data growth or sizing information is available about MPRIME and Webridge at the present time.
Therefore hardware configuration and costing estimates are just guesses with little or no factual basis.

7x24 Support not available
ITS backup and recovery architecture, and maintenance requirements prevent ITS from providing 7x24
support. ITS could commit to 7x24 with the exception of Sunday morning form midnight-noon. This
window would be needed for backups and maintenance. Note, some additional planning is needed to
assure backups can be fit into this window (due to other backups at the same time period).

Novell authentication may need to be supported
Faculty at the U-M Medical Center tend to be more familiar with their Novell authentication (NDS) than
their Umich password. The Medical Center would like to see two types of authentication available for
MPRIME. It is not clear if this is technically possible, or if it is, if it is wise to do.

Logical servers may be better than physical servers
Instead of purchasing several smaller servers, it may be better to by larger more powerful servers and
use logical partitioning to create logical servers. While this may prove a wise strategic direction,
funding issues could make this difficult to address.
Page 14 of 16
Appendix 1: MPRIME Preliminary Hardware Configuration and Pricing Information
One Time Costs
Component Name
Description
Projected Cost
New Server #1
Production
Web and Application Server
4-CPU Windows Server with X GB of RAM
3x36 (RAID5) direct-attached disk
$20,000
New Server #2
Production
Database & File Server
4-CPU Windows Server with X GB of RAM
2x18 (mirrored) direct-attached disk
SAN Adapter
$23,000
New Server #3
Development & Quality Assurance
Web and Application Server
2-CPU Windows Server with X GB of RAM
3x36 (RAID5) direct-attached disk
$12,000
New Server #4
Development & Quality Assurance
Database & File Server
2-CPU Windows Server with X GB of RAM
2x18 (mirrored) direct-attached disk
SAN Adapter
$15,000
New Server #5
Hot Spare/Upgrade
Web, Application, Database & File Server
2-CPU Windows Server with X GB of RAM
3x36 (RAID5) direct-attached disk
SAN Adapter
$15,000
Prod/Report Disk
200GB at $50 per GB
$10,000
Dev/QA SAN Disk
200 GB at $50 per GB
$10,000
Subtotal
$110,000
Ongoing Costs
Description
Projected Cost
Server hardware maintenance (15% of purchase price)
Includes SAN
$16,500
Server hardware replacement (20% of purchase price)
Includes SAN
$22,000
Disaster recovery hot site equipment
Subtotal
$1080
$39,580
Page 15 of 16
Appendix 2: MPRIME Preliminary Software Configuration and Pricing Information
One Time Costs
Component Name
Description
Projected Cost
CA Unicenter
Monitoring and job scheduling
Needed on production database/file server
Tivoli Storage Mgr
Backup and recovery
Needed on all four new servers
$3,850
MS Windows OS
Operating System
Needed on all four new servers
Included in Hardware Price
$0
MS Internet
Information Server (IIS)
Web Server
Needed on three servers, and all developer &
migrator workstations
Included in Hardware Price
$0
MS Transaction Services
Transaction Server
Needed on three servers, and all develop &
migrator workstations.
Included in Hardware Price
$0
MS SQLServer
Database software
Needed on two database servers.
Included in Hardware Price
$0
MS Visual SourceSafe
Version control software.
Free from Microsoft
$0
$13,000
Subtotal
$16,850
Ongoing Costs
Description
Projected Cost
CA Unicenter maintenance
$3,830
Tivoli Storage Mgr maintenance
$1,400
MS Windows OS maintenance
$
MS Internet Information Server (IIS) maintenance
$
MS Transaction Services maintenance
$
MS SQLServer maintenance
$
MS Visual SourceSafe maintenance
$
Subtotal
$5230
Page 16 of 16
Download