15
Thoughts on the ATF of the DataGrid 24 th –25 th May 2001 Steve Fisher / RAL

Thoughts on the ATF of the DataGrid 24 th –25 th May 2001 Steve Fisher / RAL

Embed Size (px)

Citation preview

Page 1: Thoughts on the ATF of the DataGrid 24 th –25 th May 2001 Steve Fisher / RAL

Thoughts on the ATF of the DataGrid

24th–25th May 2001

Steve Fisher / RAL

Page 2: Thoughts on the ATF of the DataGrid 24 th –25 th May 2001 Steve Fisher / RAL

24-25 May 2001DataGrid ATF - Steve Fisher/RAL 2

Aim of this presentation• Not about the ATF outputs• but about how ATF has developed• and how it functions

• Which aspects are worth copying• What we did wrong.

Page 3: Thoughts on the ATF of the DataGrid 24 th –25 th May 2001 Steve Fisher / RAL

24-25 May 2001DataGrid ATF - Steve Fisher/RAL 3

Who are the ATF• One person from each middleware WP• One (two) person(s) from WP8-10• Brian Tierney – Consultant• Ian Foster & Carl Kesselman – Consultants

– Mostly by e-mail

• Dave Kelsey – Security– Recent addition

• Fabrizio Gagliardi• Currently chaired by me

Page 4: Thoughts on the ATF of the DataGrid 24 th –25 th May 2001 Steve Fisher / RAL

24-25 May 2001DataGrid ATF - Steve Fisher/RAL 4

Why do we exist?• We have no clear mandate

• We have no defined lifetime

• For the Amsterdam meeting I invented an informal mandate.

Page 5: Thoughts on the ATF of the DataGrid 24 th –25 th May 2001 Steve Fisher / RAL

24-25 May 2001DataGrid ATF - Steve Fisher/RAL 5

Why do we exist? – Amsterdam version1. To define a viable architecture in terms of a set of components2. To ensure that these are a useful set of components able to

interwork to meet the (evolving) requirements3. To alert the PTB to components that appear not to be fully

covered by the various WPs.4. For each deliverable, collect input from the WPs and produce a

document showing the essential functionality. Inform the PTB if it appears that a component will not have the functionality required by another component.• This led to us getting mixed up in management• We are being asked to produce a dependency graph• Will we do it for M21 and M33?

5. Make proposals to the PTB on certain technical matters.• This has sometimes been controversial but is, I think, essential

Page 6: Thoughts on the ATF of the DataGrid 24 th –25 th May 2001 Steve Fisher / RAL

24-25 May 2001DataGrid ATF - Steve Fisher/RAL 6

Brief History• A bumpy start last year

– no one feeling responsible for making it work

• I started chairing the ATF in February. – The Amsterdam meeting was approaching and we had

something to deliver

• There were plans to appoint an architect – but suitable candidate not found

• First DataGrid meeting in Amsterdam • Planned to continue 1.5 days / 2 weeks

– Hard to find free days with the frequency

• Joint session with the WP8-10 TWG• Some of us participated in the ANL Globus meeting.• Technical coordinator is about to be appointed?

– Need to find right relationship with ATF

Page 7: Thoughts on the ATF of the DataGrid 24 th –25 th May 2001 Steve Fisher / RAL

24-25 May 2001DataGrid ATF - Steve Fisher/RAL 7

E-mail• We communicate extensively by e-mail.• Unfortunately a number of mail storms have been about the

mailing list policy– ATF mailing list saga was bad

– Problems with non-ATF members of the list(s)• Some complained of too many mails• Some complained that they were in the dark• Some intervened and delayed us

– Have experimented with various combinations of public and private lists

• Now we have one list with all ATF members and others who been asked to be added – but the extras are requested to send any comments via Fab or myself

– This works!!!!

Page 8: Thoughts on the ATF of the DataGrid 24 th –25 th May 2001 Steve Fisher / RAL

24-25 May 2001DataGrid ATF - Steve Fisher/RAL 8

Plan to Amsterdam• Produce model of system (UML)• Define services –

– then check if there is obvious mapping to an existing WP

• Write an architecture/design document based on Model

• Define month 9 functionality

As we proceed various issues appeared and we try to make decisions both architectural and technical

Page 9: Thoughts on the ATF of the DataGrid 24 th –25 th May 2001 Steve Fisher / RAL

24-25 May 2001DataGrid ATF - Steve Fisher/RAL 9

First architecture document - Amsterdam

• First version of a document which defines the architecture of the EU DataGrid. – This is a large undertaking – Unlike other software projects we have worked on, many

parts of the system are new (to us).– We are not just re-painting and re-assembling old

components but planning to build something new.

• Consequently for now we can only describe what we see as a feasible architecture for the DataGrid. – The new components will have to be prototyped – some will

just not work.

• The document will evolve during the lifetime of the project.

Page 10: Thoughts on the ATF of the DataGrid 24 th –25 th May 2001 Steve Fisher / RAL

24-25 May 2001DataGrid ATF - Steve Fisher/RAL 10

Life after Amsterdam

• After Amsterdam WP8-10 came back with a long list of criticisms of the architecture document.

• WP8-10 now working more closely with us to make sure that we produce something close to what is wanted

• Questions to ask WP8-10– These questions will be those we need to be

answered to help us put together the architecture.– 3 sets so far

Page 11: Thoughts on the ATF of the DataGrid 24 th –25 th May 2001 Steve Fisher / RAL

24-25 May 2001DataGrid ATF - Steve Fisher/RAL 11

Beyond M9• It seems likely that we will continue to evolve

the architecture• As things get implemented evolution

problems start – need to evaluate impact of proposed changes

• Meeting frequency and work load might reduce

Page 12: Thoughts on the ATF of the DataGrid 24 th –25 th May 2001 Steve Fisher / RAL

24-25 May 2001DataGrid ATF - Steve Fisher/RAL 12

Failures• Too many criticisms of the first architecture

document– Maybe this was inevitable.– WP8-10 had no requirements.– We wrote what looked good to us– Should have established better contact from the

beginning – but – WP8-10 presence can also slow things down as they don’t stick strictly to requirements.

• Very hard to produce UML diagrams– All agree in principle with UML

Page 13: Thoughts on the ATF of the DataGrid 24 th –25 th May 2001 Steve Fisher / RAL

24-25 May 2001DataGrid ATF - Steve Fisher/RAL 13

Successes• Identified some missing functionality and

activities• Got V1 of the architecture document out on

time

Page 14: Thoughts on the ATF of the DataGrid 24 th –25 th May 2001 Steve Fisher / RAL

24-25 May 2001DataGrid ATF - Steve Fisher/RAL 14

Bad features• No mandate• No chair from the start • The role of the experiment representative to

the ATF not clearly defined• All the middleware, experiment and security

people must attend every meeting – no redundancy

• No architect

Page 15: Thoughts on the ATF of the DataGrid 24 th –25 th May 2001 Steve Fisher / RAL

24-25 May 2001DataGrid ATF - Steve Fisher/RAL 15

Good features• No architect

– All members must have a vision

• Composition• Meeting frequency (1.5 days every 2 weeks

or so).