Dear All,
Please find attached the weekly GridPP Project Management
Board Meeting minutes. The latest minutes can be found each week in:
http://www.gridpp.ac.uk/php/pmb/minutes.php?latest
as well as being listed with other minutes at:
http://www.gridpp.ac.uk/php/pmb/minutes.php
Cheers, Tony
________________________________________________________________________
Tony Doyle, GridPP Project Leader Telephone: +44-141-330 5899
Rm 478, Kelvin Building Telefax: +44-141-330 5881
Dept of Physics and Astronomy EMail: [log in to unmask]
University of Glasgow Web: http://ppewww.ph.gla.ac.uk/~doyle
G12 8QQ, UK Video - IP: 194.36.1.33
________________________________________________________________________
GridPP PMB Minutes 233 - 23rd October 2006
==========================================
Present: Tony Doyle, Sarah Pearce, Stephen Burke, David Britton,
David Kelsey, Dave Newbold, Steve Lloyd, John Gordon, Peter Clarke,
Glenn Patrick, Neil Geddes, Suzanne Scott (Minutes)
Apologies: Roger Jones, Tony Cass, Robin Middleton, Jeremy Coles
1. Review of ACTIONS AS AT 16.10.06
====================================
225.3 The summary spreadsheet of T1-T2 networking based on GridPP3 figures
was delayed for a further week, ongoing.
228.1 JC and JG are awaiting input re the draft on dcache/DPM convergence,
ongoing.
228.2 RM had now circulated a summary on middleware areas against 3
criteria, action closed.
231.1 A video was now available re CERN footage including ATLAS film
material, action closed.
231.2 DN will send SP a draft news piece about his experiences as User
Board Chair, ongoing.
232.1 RM to provide upper-limit costs for the ATLAS Distributed Analysis
Tutorial, ongoing.
232.2 A list of R-GMA systems had now been received along with
modifications from Antony Wilson. JG will compile the document today to
be included in one of the referee responses. Action closed.
232.3 SL had now edited the PMB responses to reviewers' comments (for
reviewer 43). Action closed.
232.4 No comments regarding referee 43 had been received. Action closed.
232.5 SL had now reviewed documents 44 and 45, and extracted and answered
the questions which require a response. Action closed.
232.6 Replies from experiments had now been received and DB is working on
the proposal to the PPRP. Action closed.
232.7 A statement from ATLAS in response to Q4 (and the disparity between
CMS and ATLAS) had now been received, DN had reminded RJ and convergence
was still required. Action closed.
232.8 JG still has to draft a response to Q6 re leadership roles
incorporating a paragraph on security from DB - deadline end of today,
ongoing.
232.9 SS has provided amended documents to DB - SS and TD will draw
together any outstanding sections of the document regarding costs of posts
for GridPP 2 and 3 in response to Qs 7 and 8, ongoing.
232.10 JG has now drafted a response to Q10. Action closed.
2. Scenario Planning [DB]
==========================
DB had been awaiting experimental input regarding hardware requirements,
just received in the case of CMS. DB began with hardware costs for
GridPP3, making empirical extrapolations regarding future requirements.
He compared CPU costs and past purchases, noting that this year's purchase
was less expensive than expected. Taking a stance on future hardware
costs required a contingency line of 20-25% - no changes to tape costs.
Did the PMB want to take an aggressive stance regarding pricing for
contingency? TD noted that extrapolating on the CPU side, next year we
may pay the same price as currently. The PMB agreed the proposed stance
was realistic.
Regarding Tier-2 hardware costs, a spreadsheet was circulated by email -
power costs can be added-in. We could offer to pay Tier-2 hardware at 80%
of the cost minus power. The PMB discussed the issue of power costs and
noted a need to build-in a contingency line for power costs at Tier-2
similar to the contingency line for Tier-1. It was noted that the model
at present was that universities are not charging for power and PPARC
assume we recover this cost out of Estates or Overheads. It was felt
better to explicitly note the power cost as DB had done. It was noted
that we should reserve a sum for contingency and see what happens rather
than changing the Tier-2 figures. Costs would be monitored and if they
diverge upwards from the model presented we could request further funding.
Regarding hardware requirements, new requests had been received from
ATLAS, CMS, and LHCb including global requirements - figures presented are
all percentage of terabyte storage. It was noted that ATLAS require 12.5%
of global Tier-1 resources. The Tier-2 request is 13.33% but should be
12.5% to achieve consistency. The CMS request is 8% of global Tier-1.
The LHCb request is up from 16% to 18% based on increased UK authorship in
the collaboration. There was a discussion regarding having a defendable
baseline prior to making cuts. TD noted that the first step was to get
experiment statements and they would have to defend their request, then a
global reduction could be asserted but not associated with individual
experiments, it would be a pro-rata reduction. It was felt we should
present what the experiments say, give the global authorship numbers and
let the panel make up their own minds - all three experiments have
submitted percentage scenarios.
DB then presented his series of slides on scenario planning showing
top-down models and bottom-up worksheets along with summaries. He covered
the areas of Tier-1 and Tier-2 staff, data management, storage, workload
management, realtime monitoring, Infomon, security, and outreach. The PMB
endorsed DB's approach, noting that the key thing is to be able to propose
realistic scenarios in the 70% case - the 50% case would incur successive
losses. The PMB agreed that all of the losses should be listed and the
case put for each as being unrealistic. TD noted that no high-level
statements had yet been done - we have simply responded to the questions.
The PMB agreed that DB should prepare additional slides on what the losses
mean.
[Note: text was prepared shortly after the meeting].
3. Draft PPRP Response Document [TD]
==============================
It was noted that SL and TD had discussed and compiled a document.
Q1 - related to LHCb and ATLAS
Q2 - re funding scenarios - DB needs to summarise the experimental input,
TD and DB would iterate today and a draft will be sent to the CB.
Q3 - each of the experiments' inputs have been collated into a single response
Q4 - regarding ATLAS/CMS computing models - the statement is prepared, DB
to add to it
Q5 - re Tier-2s, SL has completed this
Q6 - leadership roles - DB and JG will provide input today
Q7 - GridPP2 extension re loss of staff - need to refer to all of the
changes since the proposal went in - TD and DB to compile
Q8 - asks for justification of each post - the 50-page document to be
appended separately will be provided to PPARC staff
Q9 - issues of QA in Tier-1 and Tier-2 - for Tier-2 a statement exists
which can be modified to relate to Tier-1 - JG will do this
Q10 - response from JG incorporated
It was agreed that the collation of responses to the referees should be a
separate document - it is around 10 pages in length. The PMB hope to get
CB endorsement of their responses to the PPRP on Wednesday.
[Note: CB endorsement and feedback was provided on Wednesday]
4. Response to Reviewer Number 46
==================================
The PMB will add further comments by tomorrow morning and email them to TD.
5. PPARC Road Map & Science Committee prioritisation
=====================================================
There had been a request to provide input to PPARC, who have a list of
projects of which GridPP is not one - GridPP doesn't exist on the project
page. Small sections should be provided under the following headings: 1.
Impact, 2. Use of subscription, 3. Track Record, 4. Momentum, 5. UK
Involvement - these are the areas by which the science section analyse
their projects. It was agreed to use the GridPP3 proposal - SP will draft
paragraphs. It was noted that the deadline had been pushed back by one
week, and this was to be done by 11th November.
STANDING ITEMS
==============
SI-1 News Items and Meeting Dates
----------------------------------
SP reported a news item on the dashboard from Martin Serozha. SP hopes to
use the GridPP infrastructure picture as picture of the week. There had
been a meeting re LHC at Home last week which was helpful and semi-technical.
Neasan O'Neill was going to CERN regarding the moving of the server.
SI-2 Production Manager's Report
---------------------------------
In absentia, JC reported the following:
1) The main issue to report is a vulnerability in PBS/Torque which some of
you may have seen reported on LCG-ROLLOUT and elsewhere. A patch from
LCG was put out late on Friday but the Tier-1 and others required the
patch for a newer version. This was released on Saturday by the
software provider. The implication of this "critical" vulnerability
was that many UKI sites either suspended their queues or put themselves
in downtime on Friday. The risk was probably quite low (it requires
someone to have a grid certificate) but the vulnerability could allow a
user to gain root on WNs and was treated as higher risk - we need a
better system to assess the risk. Sites as they patch are coming back
online.
2) Two disks on the new hardware being used under CASTOR for CSA06 have
failed and now entering rebuild which may affect the good rates we have
been seeing - CMS have been very happy with the RAL rates.
3) JC had not seen any confirmation that GridPP will be funding Tier-2
admins to attend the January WLCG service meeting to the same level as
the last Tier-2 workshop.
It was agreed that JC should collect proposals from whose who want to go
from the Tier-2s - a list is to be provided. It was noted that the
deadline for this was end November.
SI-3 LCG Management Board Report
---------------------------------
There were no issues to report. The CRRB was taking place tomorrow.
SI-4 Documentation Officer's Report
------------------------------------
SB reported that the user info group meeting will be held tomorrow.
AOB
===
SL raised the issue of the Cambridge Ontology outfit - Cambridge RE had
refused to provide a certificate because they were not part of a
University. This had eventually been agreed but may cause problems in
future for those not part of a University. It was noted that 'general
principles' regarding this issue were fine (with respect to central CA
policy) but RA have their own policies and a procedure is required. It
was noted that one was already in place. SB would deal with this issue
and inform Cambridge of the policy.
ACTIONS AS AT 23.10.06
======================
225.3 DN to provide summary spreadsheet of T1-T2 networking based on
GridPP3 figures.
228.1 Re the draft on dcache/DPM convergence, JC and JG will discuss to
decide what goes into the PPRP response.
231.2 DN to send SP a draft news piece about his experiences as User Board
Chair.
232.1 RM to provide upper-limit costs for the ATLAS Distributed Analysis
Tutorial.
232.8 JG to draft a response to Q6 re leadership roles incorporating a
paragraph on security.
232.9 DB to draw information together regarding costs of posts for GridPP
2 and 3 in response to Qs 7 and 8 - SS & TD also working on this.
233.1 DB to prepare a list of all losses and put the case for each as
being unrealistic - additional slides to be prepared on what the losses
would mean.
233.2 DB to summarise the experimental input for PPRP response document,
TD and DB to organise a draft for the CB.
233.3 DB to add to the statement for Q4 regarding ATLAS/CMS computing models.
233.4 TD and DB to compile note of changes to staff posts in GridPP2
extension - with reference to Q7
233.5 JG to amend the Tier-2 statement on QA to reflect Tier-1
requirements - in relation to Q9
233.6 JG to add comments on response to Q10
233.7 All - to provide responses to Reviewer 46 by emailing comments to TD.
233.8 SP to draft paragraphs by 11th November in response to PPARC Road
Map request for project information.
233.9 JC to collate list of those who wish to attend the January WLCG
service meeting from Tier-2.
233.10 DK to contact Cambridge regarding CA policy for issuing of
certificates.
It was noted that there was no meeting next Monday; the face-to-face
meeting would take place on Tuesday. One item for this Agenda was GridPP
members' input into the RAE, raised by PC.
The meeting closed at 4.00 pm.
|