OWG meeting 19-Jul-2007

From GlueXWiki
Revision as of 13:55, 27 August 2007 by Wolin (Talk | contribs)

(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search

Agenda

  1. Review minutes from the 15-Jun-2007 meeting
  2. Online Action Items
  3. Additions to DAQ/Online/Controls main page
  4. Need for online design goals document (200kHz trigger, 1 GB/sec, 100 Mb/sec, etc)
  5. Role of JLab DAQ group in Hall D online - Graham Heyes
  6. Preliminary Online Work Breakdown


Time/Location

10:00am Thu 17-Jul-2007 Cebaf Center F326


Announcements


Next Meeting

  • Thurs 9-Aug-2007 10am CC F326


New Action Items from this Meeting

  • Design document listing basic requirements and design goals (EJW)
  • Milestones from JLab DAQ group perspective (Graham H)
  • Milestones from Hall D perspective (EJW)


Minutes

Attendees: Elliott W, Fernando B, Vardan G, Carl T, David L, Dave A, Elke A, Simon T, Elton S, Graham H, Dave D.

Elliott went over minor changes to the Wiki pages, including the beginnings of an online task list, a list of potential future meeting topics, and a page listing the status and disposition of all action items.

Next up was a discussion of the need for a one-page document listing online design goals and requirements (thanks to Elton for pushing this). It will list the basic rate and other performance requirements, and will form the basis for understanding the roles of the various groups involved and how the work will be divided. This document will evolve as new personnel arrive, roles got modified, etc. Elliott agreed to get a first version out by the end of next week.

Graham next described his vision of the role of the JLab DAQ group in the Hall D online. This led to a long, open discussion covering many topics and points:

  • History of JLab DAQ group.
  • Development of the CODA toolkit from the early 1990's (vsn 1) until now (vsn 2, and soon vsn 3).
  • DAQ group provides software toolkit, some DAQ-specific hardware, hardware recommendations, front-end drivers, etc.
  • Experiments must provide an online group to adapt CODA to their experiment.
  • DAQ group provides support, consultation, helps with trouble shooting, etc, but experiment experts usually are the first responders when problems arise.
  • DAQ group is effectively on call 24 x 7 x 52
  • Hall B made major mods to their version of CODA, but this is discouraged. Some H-B mods will be integrated into CODA.
  • Current CODA development is very strongly motivated/directed by Hall D requirements.
  • Broadly speaking: Graham is working on the multiple-parallel event-builder. Vardan is responsible for the run control system. Carl and Elliott work on many projects involving data/message transfer and I/O, CODA system architecture, etc. Dave A handles front-end hardware and drivers. Ed works on trigger distribution, timing, designs special hardware, etc.
  • DAQ group will deliver all software needed to move data from front-ends to Unix to L3 farm to local storage. Will also deliver front-end drivers for standard hardware.
  • H-D supplies trigger, but DAQ will distribute trigger to front-ends.
  • H-D must supply software implementing the L3 trigger algorithm.
  • Graham is not sure there are currently enough people in the group to work on the L3 farm infrastructure and management software. Elke pointed out this is not needed on day 1. Elliott pointed out that some of it is needed to implement the prototype farm, which is currently in the project scope.
  • H-D needs to push on Larry, etc. to help Graham get more personnel. Design documents, milestones, timelines, and task lists will help.
  • EVIO is the default data format used by CODA. Special formats will be transported by CODA, but little more. Halls A,C,D seem happy with EVIO. Hall B is currently using BOS/Fpack, but very likely will switch to EVIO for the upgrade.
  • Dave D asked about the DAQ group's ability to simulate large systems. Graham says they have collected many retired nodes from the QCD and batch farms, so this is possible and has been done for some components. No full high-speed system tests have been performed yet.
  • Elke pointed out that the timescale for such demonstrations is Jan 2011.


Concerning more of the hardware aspects of the DAQ group effort:

  • Ed designs most DAQ-specific hardware.
  • Design of the interface between the trigger, front-ends, and CODA software needs to start soon. It will be used in all four halls for the upgrade.
  • Halls A and C are much simpler than Halls B and D. Whatever works for H-D should work for the others.
  • Some trigger/timing hardware is done via a collaboration between the DAQ and Electronics groups.
  • The Electronics group manpower is ramping up. DAQ group is not.
  • The new Trigger Interface (TI) card is high on the list, should be coming out next. It will support event blocking, will allow testing of crates of F1TDC' s and FADC's, etc.


Concerning the formal relationship between DAQ and Hall D and CODA 3 development:

  • Need roadmap with milestones for CODA development.
  • H-D needs to influence priorities at the lab (manpower, etc).


Finally, some action items:

  • Graham will produce list of CODA 3 milestones, without dates, in the next week or so.
  • Elliott will do the same from the Hall D perspective.

These will be sent around for comment, then a mutually-consistent set will be developed, then dates will be added.