[CFIC] CFIC: Meeting information

Pat Waukau waukau at ucar.edu
Thu Jan 24 11:43:11 MST 2008


CFIC Members:

Following is the agenda for today's meeting. It is also on the wiki site.
In addition, I am including a proposal developed by Lance for discussion at 
today's meeting.  

Agenda - CFIC Meeting - January 24, 2008, 2:00 PM, CG4-2020

   1. Intro/Check In
   2. Group Updates
          * Governance and Management
          * Organizational Processes
          * Site Studies
          * Cost Allocation
          * Transition
   3. Open Discussion
   4. Proposed Timeline
          * 2/29/2008 - Draft report for committee review
          * 3/6/2008 - CFIC meeting
          * Distribute report to and meet with system adminstrators
          * 3/21/2008 - PC Presentation
          * 5/8/2008 - UMC Presentation


Lance's Proposal:

Very briefly here is the "alternate" plan (and it should look familiar
since we've talked about it before and it's not radically different than
what we've been talking about).  We can think of this as the
"consolidation" or "transitional"  strategy.

First we define two classes of server rooms: production (ML29, FL2-3095,
CG2) and development labs (some to-be-determined subset of existing server
rooms).  Production server rooms are centrally managed by the Management
group as defined in the working group report.  Development labs continued
to be managed by the divisions or labs.  There are different SLAs for each
class of room (i.e. power and cooling failures in development labs are not
responded to as quickly as production rooms).

We need to look at the existing set of development rooms and come up with
a proposed plan for how to consolidate rooms.  In other words, we're
looking to retire approximately half of the existing development rooms-
preference to be given to shutting down rooms on building air, and on
those groups who are behind moving to new facilities.

We can use UOP as our example of how this all could work.  I've been
talking to UOP management and sysadmins and all are supportive of this
idea.

I think this plan is superior to the existing ML29-centered plan for
several reasons.

1) we can get broad support for this from the sysadmin community
2) the upfront costs are potentially lower than the ML29-centered option
3) we don't need to wait until 2009 (or later... or never) to get rolling
on this- we can start saving money now.
4) there's something in this plan that all parties can view as a win for
their interests
5) it doesn't require a major change in direction from what the committee
has been pursuing up to this point
6) if we find that the benefits aren't sufficiently great, we can still
move toward further centralization later on with minimal penalty
7) we can finally wrap this up and deliver our final report on schedule in
March



-- 
Pat Waukau   waukau at ucar.edu
Head, Systems Group
Mesoscale and Microscale Meteorology Division
NCAR
P. O. Box 3000 
Boulder, CO  80307
(303) 497-8906 (Office)
(303) 497-8181 (FAX)



More information about the CFIC mailing list