Minutes of the JDCMG Meeting

October 26 and 27, 2004, UCB

Agenda

The meeting started with introductions. Attendance included representatives from UCB, UCB-Lab, UCD, UCI, UCLA, UCOP, UCSB, UCSF Med. Center, and UCSD.

Vendor Scorecards:

A common vendor scorecard across all campuses has a lot of support but it was agreed the place to start with an expanded share inventory. We agreed to develop a shared spreadsheet that would list the core key vendors and ask each campus to provide the following information:

We also would have links to Gartner information about each vendor including company profile with financial viability, market position, market trend, environmentally aware, support of minorities, etc.

In addition to the above information, we will have a drop down high, medium, low rating for each campus to provide feedback on level of satisfaction for the vendors:

• Products
• Service
• Pricing

There was a concern that there would be objections in purchasing departments so we each a greed to send a note to our purchasing people that we were doing it. We reached Agreement that this is not to be shared with the vendors.

The vendors we want to include as starting are:

Shel will write a template for the above and include a process so each campus can add vendors. Charlotte will get it added to the Web site with the capability to have members only update it.

Recharge:

UCLA has been doing recharge based on full cost accounting. They have done a "tax" for things like telephones in the past. They had previously done a full cost accounting but it was never accepted. Don Worth is now writing a proposal to do "technology tax" to fund key pieces of shared infrastructure. Below is a recap of some of their process/ procedures:

• Recharge: Data Center services big sell is 24 x 7 support. Often departments have only one person available to support environments and that isn't sustainable.
• Colocation @ UCLA: don't want give departments access and they strongly want them to use central sys adm services. Basically colocation provided includes system admin.
• They have developed "Powerunits", simple server, 1 powerunit. Small database 2 powerunits, Data Warehouse is 8 units, $600/month per powerunit (space and sysadmin space). Includes backup services, sys adm, network, and rack units. They are doing it for about 128 application servers (internal AIS department servers). 15-20 application develop servers from outside. Staff of 7 distributed server Sys Admins. They put everything on the same network. If the customer is going to do the sys admin, we will likely put them on their own VLAN. Hardware is always purchased, server and disks, by the department that owns the applications.
• They use a time sheet applications they developed in house to show how much effort per application. They keep track approximately how many hours they put into servers. They spend about 30-40 hours per powerunit per year. They modify their rates every fiscal year. Its always been within 10%, it has gone down more than its gone up. Much of their funding is based on permanent allocation for service

UCSF: Mark indicated they have a very high regulation requirements for HIPPA compliance which includes server compliance so departmental servers are likely. In the most recent times they are doing just their first proposal. The customer is purchasing a new server and wants full service on it. Startup costs including rack space, installation, software licenses for the server (OS, security), network connections (on the data center network), tape backup, for $3500 startup and monthly $395. $12.50/month per rack unit. $110/CPU/Month sys adm, $25 per network connection, $25 backup.

DAVIS: Windows servers support requires more support staff than UNIX on the order of 3 windows admins for 30 and 7 unix admins for 150. Unix is primarily solaris, some HPUX and a little bit of linux. The recharge rates are $67/hour for sysadm and database admin. They propose on expected time but only bill on actual time used. For the actual housing, it is by rack unit per month. They originally charged on a cost recovery over three years but have gone to charging by monthly units. They require that customers have hardware maintenance costs. They calculate based on actual tape used (based on nightly storage costs). That encourages them to end of life retention sets. They track all the time spent based on an internal home grown application to track time. They do not include staff costs for things like backups, operations, etc as those are core funding.

UCSB: Total budget is $3M. $1.6M total recharge. 80% of that is centrally funded. The rest is from small 10k increments for many server kinds of recharges. They have changed platforms (added unix and windows servers) out of the mainframe cost center. They have done it buying used equipment and stretching staff costs. The model discourages the use of the central resources and encourage the use of things that are not on the central recharge so local departments are creating their own shadow organizations. Their collocation with full system administration is $300/month. That includes a 1/10th of an FTE. They have some unattended operations between 9am and 6am.

Service Level Agreements: We did not get to these this meeting and they will move forward to future discussion.

Next Meeting:

• Conference Call in January 12th (Wednesday)
• Vendor Scorecard Demo
• Service Level Agreements
• Next Face to Face Meeting is March 04 at UCI
  • Disaster Recovery Issue