Grand Challenge Update & News: 8-Sep-1997

by D. Olson

Topics:

Results of RHIC Computing Review (29-31 July 1997)

RCF Meeting notes (7-8 Aug 1997)

RHIC Event Store Task Force

GC Workshop at BNL Oct. 2-3, 1997

RHIC Computing Review:

As many of you know, there was a DOE review of RHIC Computing at Brookhaven on the 29-31 July 1997. The scope of that review covered all aspects of off-line computing for RHIC including:

There were numerous recommendations by a very competent review committee and I will only mention the ones with direct impact on the grand challenge project. My paraphrasing (not quotes from the report) of these are:

  1. the experiments should review and prepare a more thorough set of requirements with an emphasis on understanding the characteristics of the processing needs (I/O - CPU ratio for example)
  2. RHIC should decide very quickly on the implementation for the event data storage system
  3. the grand challenge project should focus on what direct contribution it can make to RHIC computing

The implication of recommendation 1 is relevent to the hardware environment for data analysis and will affect the environment for developing software for the grand challenge.

The result of recomendation 2 was the formation of a task force which includes members from RHIC Computing, the RHIC experiments and the grand challenge project. This is discussed in more detail below.

The result of recommendation 3 is that the efforts in our grand challenge project be directed at (and only at) items which will be delivered to RHIC computing.

RCF Meeting 7-8 Aug 1997

At the bi-monthly RHIC computing meeting on 7-8 Aug 1997 there was an afternoon session for the grand challenge with the aim of better defining the set of components the GC project can be responsible for in delivering to RHIC.

Part of this discussion was about the analysis architecture and a revised version of this diagram generated (96 dpi gif, 72 dpi gif, Windows Draw). The major difference was the separation of the "job control" and the "query formatter" (now called "query estimator"). Also, a "utilities" component which permits copying data without an analysis process is shown explicitly as well as a "filter" which appeared in earlier drawings.

Items identified in this discussion (not yet a complete list) are:

  1. The GC will participate in evaluating Objectivity (C.T., D. M.) (RHIC data store working group)
  2. A.S. will provide disk-based object (event) index & query processing
  3. The parallel processing addition to STAF & job control will be provided by the GC (person is TBD)
  4. load balancing & coherent access will be provided by RCF
  5. disk cache management is tentatively assigned to A.S.

RHIC Event Store Task Force

Following the recommendations of the RHIC computing review committee a task force was set up in order to recommend and event storage system or strategy, by Nov. 1, 1997. Information about this task force is available on the RHIC web at http://www.rhic.bnl.gov/html/event_store_task_force/event_store.html.

This committee is comprised of members of RHIC Computing, the RHIC experiments and the grand challenge project. The activities of this task force are extremely important as it is defining requirements and software products which we will work with in the grand challenge project.

Grand Challenge Workshop at BNL, Oct. 2-3, 1997

The next in-person meeting for the grand challenge is scheduled for Oct. 2-3, 1997 at Brookhaven National Lab. It is a 2-day workshop in between a RHIC computing meeting (10/1) and a RHIC common off-line software meeting (10/4).

As the name implies, it will be a workshop format with some status reports, no doubt a lot of discussion about the RHIC event store & Objectivity, and serious discussion of the deliverables and schedule which the grand challenge can provide to RHIC.

Updates:

11 Sept. 97, changes links to architecture diagram. D. O.