From: Albert Lazzarini Date: Sat Jan 12, 2002 02:45:00 US/Pacific To: Albert Lazzarini Subject: Fwd: MOU with UWM for the next period of time Begin forwarded message: From: Albert Lazzarini Date: Sat Dec 29, 2001 08:38:12 US/Pacific To: Bruce Allen Cc: Alan Wiseman , Barry Barish , weiss@ligo.mit.edu, "Gary H. Sanders" , Albert Lazzarini Subject: Re: MOU with UWM for the next period of time Bruce, This is a reasonable suggestion and is consistent with the first tasks with which I want to charge our committee. Albert On Saturday, December 29, 2001, at 03:34 , Bruce Allen wrote: Hi Albert, I apologize for the delay in replying to your comments on the proposed UWM MOU for Aug 15 2001 - Feb 15, 2002. I agree with you that we need to make the Grid research committments and responsibilties of the different LSC groups as stated in the MOUs clearer and less vague. As you said, other projects like CMS and Atlas aim to be more explicit and the LIGO review committees will probably want to see these committments and plans spelled out more explicitly. Since the grid work needs to be coordinated among the LSC as a whole, and since the UWM committments should be consistent with those made by the other LSC institutions that are getting grid funding, I wanted to suggest that the wording of the different LSC MOUs get worked out by the LSC Computing Committee that Rai recently formed, and which you chair. Since the period covered by the MOU that you commented on runs for only another six weeks (till mid-February) I suggest that for this current period we let the wording stand, but then starting with the next MOU for the period Feb 15 - Aug 15 2002, we have a more precise and explicit set of committments in the different LSC MOUs, with wording to be worked out by the LSC Computing Committee. Would this be OK? Bruce On Saturday, December 1, 2001, at 02:45 , Albert Lazzarini wrote: Hi Bruce, I am sorry to have missed the chance to see you and to talk with you during the PAC meeting this past week. I am sure that your PAC presentation was well received. I had a prior commitment at FNAL all week. However, I have had the opportunity to speak to Barry and to get a brief summary of the meeting, so I am aware of some of the issues and findings from the PAC. I have reviewed the current draft of the MOU attachment with UWM for the next period of performance. The draft identifies the key components of the UWM research program and is close to a final version. However, in reading items (h) and (k), which deal with Grid research and with the UWM cluster, I found the wording to be somewhat vague with regard to the commitment being made by the UWM group to the Laboratory regarding how these activities will proceed. Having just returned from a week long review of LHC computing in the US, I was sensitized to the issues facing these other collaborations that are caused by multiple, competing, and in some regards incompatible activities that are going on within the collaborations at both the national and international levels with regard to grid computing and data analysis frameworks and architectures (US vs EU, ATLAS vs CMS, ATLAS vs ATLAS, etc.). The committee expressed unanimous concern on the risks involved in not maintaining a uniformity of approach to computing, urging for a more centralized oversight of these activities by the project management of the Tier 1 centers at the national laboratories. Although I was on the "giving end" of this review, the same concerns can (and are likely) be applied to us in a future review. We all need to plan to avoid this to the greatest extent possible. The whole NSF-supported physics computing community has made a Faustian bargain with the grid computing community in order to acquire NSF resources for computing that would otherwise not have been possible. The concern is the tension between deliverable, robust products for science vs. R&D for computer science. The review committee came down rather hard on the LHC collaborations for not fully having their act together in this regard. In many ways, but because we are smaller, we have most of the same issues on a smaller scale. Whereas the LHC experiments still have 5 years to get their act together, we face the beginning of the LIGO science run era in less than one year. It is clear that the funding agencies expect the HEP project management of the two experiments to deal with these risks by carefully tracking and to the greatest extent possible, controlling the course of evolution that is permitted. I heard from Barry that you expressed a similar concern (if not similar solution) at the PAC meeting. With these things in mind, before the UWM MOU for the next period is signed, I am recommending to Barry that the Laboratory ask UWM to enunciate clearly, in unambiguous and specific terms in the MOU the commitment UWM makes to operating LDAS as a continuous activity that is kept current with Laboratory development and on which LIGO and the rest of the collaboration can rely as a Tier 2 center. With regard to the substantial GriPhyN and iVDGL actvities at UWM, it is likewise important to better specify how the multiple options and directions in which things can evolve are going to be inter-related, controlled, limited and managed at UWM in order to prevent long term fragmentation of effort and incompatibilities to evolve. Examples of things the MOU should contain are how the UWM plans to integrate the Condor-only model vis-a-vis the grid-enabled LDAS model which was shown at SC2001, vis-a-vis the Cactus and/or Triana model of AEI? What percent of UWM resources will be devoted to these (or other activities)? What are their priorities? etc. I was able to see the MOUs the LHC project management at the Tier 1 center has been developing with its Tier 2 computing universities, and they resemble the statements of work (SOWs) the Laboratory has used with contractors. The reasons this is possible for the LHC experiments is that the project management controls the purse strings on which and how much universities receive to pursue their LHC research plans. The LIGO-LSC model is different in this respect, of course, so it is not reasonable to expect such tight "contractual" SOWs to appear in our MOUs. However, our LSC MOUs have purposely been left ambiguous in the past as we waited for the collaboration to mature. We should now start to move away from this model. As another measure that needs to be taken, I agree with you that the collaboration needs to act soon to develop a body that can act responsibly to oversee and use the NSF-provided resources. Sam has been asking for the same thing, and Barry, Rai, and I have been in email discussion about this. I am hoping that Rai will soon be in a position to act. I look forward to your revised MOU attachment. Thanks, Albert PS: I copy Barry on this in case he (or Gray) wish to provide guidance or feedback to us in this. ------------------------------------- LIGO Laboratory at Caltech Data and Computing M/C 18-34 Room 609 Millikan Library 1200 E. California Blvd. Pasadena, CA 91125 Office: +1.626.395.8444 Facsimile: +1.626.304.9834 -------------------------------------