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/pmb/minutes/this-week.txt
(as well as being listed with other minutes e.g. 021121.txt)
__________________________________________________________________
David Britton, GridPP Project Manager Tel: +44 (0)20 7594 7821
HEP Group, Blackett Lab Fax: +44 (0)20 7823 8830
Prince Consort Road Email: [log in to unmask]
London, SW7 2BW
__________________________________________________________________
GridPP PMB Minutes 58 - 21 November 2002
===========================================
Physical Meeting in Bristol
Present: Dave Britton, Pete Clarke, Neil Geddes, Robin Middleton, Nick Brook, Tony
Doyle, Steve Lloyd, John Gordon.
By video: Tony Cass, Roger Barlow.
Top-Down Plan
DB presented an intermediate version of the Top-Down plan financial plan for the
GridPP-2 era. It now included staff costs discussed with Dave Kelsey for CLRC and a
revised estimate for university staff costs that included a step increase above inflation.
NG commented that this was wise because of the possibility of a change in the way
overheads to universities were calculated. The profile of CERN posts had also been
adjusted to smooth the transition from GridPP to GridPP-2.
NG reported that his major goal was to convince Council that computing for LHC is
totally funded out of the new money. There is pressure to define the total cost of the LHC
experiments and IH had originally wanted separate bids but had persuaded that this was
not the optimal approach. NG suggested that the core bit of a GridPP-2 could go to the
Grid Steering Committee and experiment bids for application support to the Science
Committee (along with non-HEP bids). It was decided to define two budget lines:
Application Integration and Application Development. The former being essentially the
"Gridification" of applications, and the latter being the non-grid core software
development. Application integration should clearly be part of a GridPP-2 project but the
application development may be better managed by the experiments and may or may not
come under the GridPP-2 umbrella.
TC noted that CERN would like to know where the UK Tier-1 centre would be located
and what the level of support would be. NG noted that a bid for a Tier-1 implied a
commitment for longer than the next three years. A discussion ensued on exactly which
staff were included in the Tier-1 line. Although this is not true at present, it was felt that
middleware support staff should eventually be included in the Tier-1/2 numbers. It was
clear that the bottom-up proposal currently being developed would eventually lead to a
clear understanding of what was included in which line.
Bottom-up Planning
NB presented input from the experiments for the bottom-up plan. People had been
reluctant to provide numbers and there would need to be further iteration and
clarification. Nevertheless, a total requirement of about 130 FTE was identified. This
included academic effort and the net requirement for eSc funded post in the GridPP-2 era
was about 43 FTE. This includes the continuation of the 13 GridPP application posts and
the 5 WP8 posts. (No response had been received from CDF). DB felt this was an
encouraging number and proposed building this into the next version of the Top-Down
plan as the Application Development line noted above.
NG noted that we would need to give a value for each experiment from GridPP by
dividing up the common lines. Costing the each experiment separately and adding them
up should give a bigger net total and illustrate the advantage of the integrated approach.
It was agreed to standardize on SpecInts. NB and JG to get together to sort out numbers
to give to TD for the SC meeting.
EGEE and future plans.
RM reported on EGEE and GridPP. This is very much a work in progress. The question
is how much we buy in to EGEE and what we want to do on our own in the UK. EGEE
links to all e-Science, EU and US/AP. RM outlined a possible general approach of
professional full time management (100%) of work areas. We could focus on small
number of core centres with particular strengths. He asked whether GridPP should define
work deliverables/MoU with core centres. NG said this is also an interesting question
with regard to Tier-2 centres. TD argued against it as the deliverables get too tightly
defined. We probably need agreements for hardware and support but not for
development. RM presented a matrix of current middleware strengths. For the future he
suggested data management (ScotGrid) (+ mass storage). Networking (London and
NW?), Information Services (CLRC). It was suggested that we move out of fabric
development (but obviously not deployment) and mass storage development. Other
possibilities are RB, /grid, Security, and CA. PC said we have identified areas above
critical mass. NG agreed that joint projects with e-Science would be good.
RM summarised EGGE. It is for infrastructure with PP only part of the project. Initially it
would be €25-35M over 2 years of which the UK share might be 10-15%. There will be a
call in mid-Dec 02, closing in Mar/Apr 03. There will be a carve up on/before 10 Dec.
TD asked where DataGrid2 is. NG said there is no call at the moment. We estimate
EGEE could result in 10-15 people in UK. TD asked about the management line and
whether there are UK people willing to put themselves forward. NG said that EPCC may
be interested and maybe PPARC/CLRC. DB said that it looks like £2M over 2 years and
asked how much matching funds there would have to be. NG said that it is though to be
50:50 though not necessary in the same areas. Our contribution could be the e-Science
Grid and Tier centres. It was agreed that not all this effort should be subtracted from our
requirements. TC asked if there will there be a lag between UK and EGEE as there was
with DataGrid. It was thought probably not. He also asked if we could it buy in support
from companies. NG said this is not planned but is not precluded. PC said
countries/centres could take responsibility for certain areas e.g. information service. TD
pointed out that the released EDG software does not include R-GMA.
Possible UK areas are Networking (liase with GEANT), Information and Monitoring,
and Suppport/Operations. TD suggested it was clear that we should be pushing
Networking and Information and Monitoring. To some extent it depends on level of
commitment required. NG said the Core Programme is interested in Support/Operations,
OGSA-DAI, Industry and Training. PC said Networking involves drawing up a work
package including, for instance, bookable bandwidth and end to end performance. Fab
assumes we are doing this. It was agreed that we need to go ahead for many reasons even
though direct financial benefit might be small. TD pointed out that PPARC's e-Science
programme is strong in Europe and we need to continue this. NG said the priority order is
probably Networking and Support/Operations above Information/Monitoring. We agreed
we have to try and be involved and recognise that it might not contribute directly to
GridPP2. DB said he would add the staff above the line and subtract again below. TD
said we agreed before that it should be 4 posts to Tier-1 and 4 to Tier-2.
PC suggested that middleware posts be concentrated. All centres should be asked what
areas they would like to concentrate on. TD said he thought the TB should be asked but
this was thought to be to narrow.
Risk Assessment
DB explained the risk document. He said that he was not looking to extend the list but
just agree the entries. The GridPP risks are those which not explicitly elsewhere. Each
risk was discussed in turn and appropriate changes made by DB. RB agreed to do the risk
assessment for the applications. Most apply to all experiments. He should consider if
there are experiment specific risks. PC argued that CDF not adopting SAM was not a risk
and it was agreed to delete this. DB said that the risk assessment documents have to be
filled out in the next 7 days by 28 November.
Science Committee
NG said that he had been asked to present what GridPP and AstroGrid are doing.
Afterwards NG will present what PPARC are going to do with the new money. The
money will to go to the Science Committee not the e-Science Steering Committee. The
money is still effectively ring fenced. The suggestion is to ask for proposals for
continuing LHC Grid work and their applications i.e. to fund all of LHC computing till
2007. There is also VISTA plus other Astronomy data centres. IH prefers that the e-
Science Committee does the call but wider Science community (PPRP) does Peer
Review. The call will probably be in February for a deadline in April maybe. SL
requested that the call be made before the next CB meeting on 19 February if possible.
NG showed the slides he proposed showing to the SC. TD started to show his SC slides
and SL, RM, DB and JG left!.
Next Meetings
There will be no PMB VC on 25 November. The next face to face will be at IC on 20
January 2002 at 9.30 for 10.
ACTIONS AS AT 21th November 2002
================================
Standing Items
==============
SI-1 : ALL : Continual review/improvement/update of the GridPP website.
SI-2 : ALL : to provide any new items to SL (who will pass on to PPARC).
SI-3 : ALL : supply SL/TD with meeting dates for the web page.
SI-4 : EB & TB to improve their areas of the web.
Actions
========
ACTION 35.4 : TD to contact sites which do not yet have a "green dot".
ACTION 37.3 : JG to review deputy job description.
ACTION 40.36: DB check and pull together risk assessments.
ACTION 53.1 : RB to generate financial requirements for Babar.
ACTION 53.3 : ALL to provide input to long-term planning document by
November PMB meeting.
ACTION 56.1 : SL to put the Long Term Planning information together after
the PMB meeting on 21st.
ACTION 56.2 : All to provide TD with extra info for the Science Committee
talk.
ACTION 57.2 : TD to provide draft Science Committee talk to the board.
ACTION 57.3 : RM to email UK HEP Grid about the PPARC ruling on travel.
|
