1 / 46

LCG Service Challenges Plan Jamie Shiers, CERN-IT-GD 22 March 2005

LCG Service Challenges Plan Jamie Shiers, CERN-IT-GD 22 March 2005. Introduction. Draws heavily on draft Service Challenge chapter for the LCG TDR And recent presentations by various people (POB, GDB, SC, …) Some high level (but worrying…) comments on SC1 and SC2

rossa
Download Presentation

LCG Service Challenges Plan Jamie Shiers, CERN-IT-GD 22 March 2005

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. LCG Service Challenges Plan Jamie Shiers, CERN-IT-GD 22 March 2005

  2. Introduction • Draws heavily on draft Service Challenge chapter for the LCG TDR • And recent presentations by various people (POB, GDB, SC, …) • Some high level (but worrying…) comments on SC1 and SC2 • Main focus is planning for future challenges (mainly SC3…) • Main difference wrt previous presentations is emphasis on different aspects of the challenges • i.e. not just raw transfer rates… • Draft milestones presented at February SC meeting well understood • Planning is an iterative, continuous process…

  3. Key Principles • Service challenges results in a series of services that exist in parallel with baseline production service • Rapidly and successively approach production needs of LHC • Initial focus: core (data management) services • Swiftly expand out to cover full spectrum of production and analysis chain • Must be as realistic as possible, including end-end testing of key experiment use-cases over extended periods with recovery from glitches and longer-term outages • Necessary resources and commitment pre-requisite to success! • Should not be under-estimated!

  4. Overall Schedule

  5. SC1 / SC2 • SC1 did not successfully complete its goals • Dec04 - Service Challenge I complete • mass store (disk) - mass store (disk) • 3 T1s (Lyon, Amsterdam, Chicago) (others also participated…) • 500 MB/sec (individually and aggregate) • 2 weeks sustained • Software; GridFTP plus some scripts • We did not meet the milestone of 500MB/s for 2 weeks • We need to do these challenges to see what actually goes wrong • A lot of things do, and did, go wrong • We need better test plans for validating the infrastructure before the challenges (network throughput, disk speeds, etc…) • SC2: Mar05 - Service Challenge II should be complete • Software: reliable file transfer service • mass store (disk) - mass store (disk), • BNL, CNAF, FNAL, FZK, IN2P3, NIKHEF, RAL (more than originally planned) • 100MB/s per T1, 500MB/s aggregate out of CERN, push 2+ sites to ~500MB/s • 1 month sustained

  6. SC2 – Current Status

  7. SC1/2 - Conclusions • Setting up the infrastructure and achieving reliable transfers, even at much lower data rates than needed for LHC, is complex and requires a lot of technical work + coordination • Even within one site – people are working very hard & are stressed. Stressed people do not work at their best. Far from clear how this scales to SC3/SC4, let alone to LHC production phase • Compound this with the multi-site / multi-partner issue, together with time zones etc and you have a large “non-technical” component to an already tough problem (example of technical problem follows…) • But… the end point is fixed (time + functionality) • We should be careful not to over-complicate the problem or potential solutions • And not forget there is still a humungous amount to do… • (much much more than we’ve done…)

  8. SC2 – Sample Network Problem • Q: Since Tuesday 15 of March at 1pm (more or less) we are experiencing very low performance for data stream originated at Karlsruhe (192.108.46.0/24) and at CNAF (192.135.23.0/24) and reaching CERN (192.16.160.0/24) via the backup 10G connection: data flow that were running at 1Gbps now are at 5-10Mbps from Karlsruhe and ~500Mbps from CNAF. There is no problem on the other way (from CERN to the two locations). (Reported March 17) • A: Following tests between our NOC and DFN, a faulty connection on the patch panel has been detected as the origin of these errors. It has been fixed and no errors appear now. (Solved March 21) • Who / how is this going to be debugged (and fixed) during Service Phase at 00:15 Sunday morning when expt framework detects something is wrong?

  9. SC2 – “Production problems” • For well understood reasons, we are currently trying to move away from ‘R&D’ style h/w (machines + network) • We need a stop-gap solution for SC2 and another for SC3 • We are along way from “production mentality” • Don’t play with things whilst we’re running production unless change request signed in blood. • All these issues can – and will – be solved but doesn’t any feel a case of Deju Vu?

  10. (Some) Meetings • SC2 Meetings • Phone Conference Monday Afternoon • Daily call with different sites • Post SC2 Meetings • Tuesday 4pm (starting April) • Storage Management Workshop • 5-7 April @ CERN • T0/T1 network meeting • 8 April in NIKHEF/SARA • Pre-SC3 Meeting • June 13th – 15th at CERN • Taipei SC Meeting • 26th April • Hepix @FZK • 9-13 May • There are also others: • Weekly SC meetings at CERN; • T2 discussions with UK • next meeting: June in Glasgow • T2 INFN workshop in Bari • 26-27 May • T1 site visits • PIC / BNL / FNAL / Triumf / ASCC • etc. etc. IMHO: meetings are needed but not everyone needs to go to all. Nor do we need same style. I want “stand-up” or “face-to-face”

  11. SC3 Preparation Workshop • This (proposed) workshop will focus on very detailed technical planning for the whole SC3 exercise. • It is intended to be as interactive as possible, i.e. not presentations to an audience largely in a different (wireless) world. • There will be sessions devoted to specific experiment issues, Tier1 issues, Tier2 issues as well as the general service infrastructure. • Planning for SC3 has already started and will continue prior to the workshop. • This is an opportunity to get together to iron out concerns and issues that cannot easily be solved by e-mail, phone conferences and/or other meetings prior to the workshop. • Is there a better way to do it? Better time?

  12. SC3 on • SC3 is significantly more complex than previous challenges • It includes experiments s/w, additional m/w, Tier2s etc • Proving we can transfer dummy files from A-B proves nothing • Obviously need to show that basic infrastructure works… • Preparation for SC3 includes: • Understanding experiments’ Computing Models • Agreeing involvement of experiments’ production teams • Visiting all (involved) Tier1s (multiple times) • Preparing for the involvement of 50-100 Tier2s • Short of resources at all levels: • “Managerial” – discussing with experiments and Tier1s (visiting) • “Organizational” – milestones, meetings, workshops, … • “Technical” – preparing challenges and running CERN end – 24 x 7 ???

  13. 2005 2006 2007 2008 SC2 SC3 First physics cosmics First beams Full physics run 2005 Q1 - SC3 preparation Prepare for the next service challenge (SC3) -- in parallel with SC2 (reliable file transfer) – Build up 1 GByte/s challengefacility at CERN • The current 500 MByte/s facility used for SC2 will become the testbed from April onwards (10 ftp servers, 10 disk servers, network equipment) Build up infrastructure at each external centre • Average capability ~150 MB/sec at a Tier-1 (to be agreed with each T-1) Further develop reliable transfer framework software • Include catalogues, include VO’s disk-network-disk bandwidths

  14. 2005 2006 2007 2008 SC2 SC3 First physics cosmics First beams Full physics run 2005 Q2-3 - SC3 challenge SC3 - 50% service infrastructure • Same T1s as in SC2 (Fermi, NIKHEF/SARA, GridKa, RAL, CNAF, CCIN2P3) • Add at least two T2s • “50%” means approximately 50% of the nominal rate of ATLAS+CMS Using the 1 GByte/s challenge facility at CERN - • Disk at T0 to tape at all T1 sites at 60 Mbyte/s • Data recording at T0 from same disk buffers • Moderate traffic disk-disk between T1s and T2s Use ATLAS and CMS files, reconstruction, ESD skimming codes (numbers to be worked out when the models are published) Goal - 1 month sustained service in July • 500 MBytes/s aggregate at CERN, 60 MBytes/s at each T1 •  end-to-end data flow peaks at least a factor of two at T1s •  network bandwidth peaks ?? tape-network-disk bandwidths

  15. 2005 2006 2007 2008 SC2 SC3 First physics cosmics First beams Full physics run 2005 Q2-3 - SC3 additional centres In parallel with SC3 prepare additional centres using the 500 MByte/s test facility • Test Taipei, Vancouver, Brookhaven, additional Tier-2s Further develop framework software • Catalogues, VO’s, use experiment specific solutions

  16. 2005 2006 2007 2008 SC2 SC3 First physics cosmics First beams Full physics run 2005 Sep-Dec - SC3 Service 50% Computing Model Validation Period The service exercised in SC3 is made available to experiments as a stable, permanent service for computing model tests Additional sites are added as they come up to speed End-to-end sustained data rates – • 500 Mbytes/s at CERN (aggregate) • 60 Mbytes/s at Tier-1s • Modest Tier-2 traffic

  17. SC3 – Milestone Decomposition • File transfer goals: • Build up disk – disk transfer speeds to 150MB/s • SC2 was 100MB/s – agreed by site • Include tape – transfer speeds of 60MB/s • Tier1 goals: • Bring in additional Tier1 sites wrt SC2 • PIC and Nordic most likely added later: SC4? • Tier2 goals: • Start to bring Tier2 sites into challenge • Agree services T2s offer / require • On-going plan (more later) to address this via GridPP, INFN etc. • Experiment goals: • Address main offline use cases except those related to analysis • i.e. real data flow out of T0-T1-T2; simulation in from T2-T1 • Service goals: • Include CPU (to generate files) and storage • Start to add additional components • Catalogs, VOs, experiment-specific solutions etc, 3D involvement, … • Choice of software components, validation, fallback, …

  18. SC3 – Milestone Decomposition • File transfer goals: • Build up disk – disk transfer speeds to 150MB/s • SC2 was 100MB/s – agreed by site • Include tape – transfer speeds of 60MB/s • Tier1 goals: • Bring in additional Tier1 sites wrt SC2 • PIC and Nordic most likely added later: SC4? • Tier2 goals: • Start to bring Tier2 sites into challenge • Agree services T2s offer / require • On-going plan (more later) to address this via GridPP, INFN etc. • Experiment goals: • Address main offline use cases except those related to analysis • i.e. real data flow out of T0-T1-T2; simulation in from T2-T1 • Service goals: • Include CPU (to generate files) and storage • Start to add additional components • Catalogs, VOs, experiment-specific solutions etc, 3D involvement, … • Choice of software components, validation, fallback, …

  19. SC3 – Experiment Goals • Meetings on-going to discuss goals of SC3 and experiment involvement • Focus on: • First demonstrate robust infrastructure; • Add ‘simulated’ experiment-specific usage patterns; • Add experiment-specific components; • Run experiments offline frameworks but don’t preserve data; • Exercise primary Use Cases except analysis (SC4) • Service phase: data is preserved… • Has significant implications on resources beyond file transfer services • Storage; CPU; Network… Both at CERN and participating sites (T1/T2) • May have different partners for experiment-specific tests (e.g. not all T1s) • In effect, experiments’ usage of SC during service phase = data challenge • Must be exceedingly clear on goals / responsibilities during each phase!

  20. SC3 – Experiment Involvement Cont. • Regular discussions with experiments have started • ATLAS: at DM meetings • ALICE+CMS: every ~2 weeks • LHCb: no regular slot yet, but discussions started… • Anticipate to start first with ALICE and CMS (exactly when TDB) ATLAS and LHCb around October • T2 sites being identified in common with these experiments • More later… • List of experiment-specific components and the sites where they need to be deployed being drawn up • Need this on April timeframe for adequate preparation & testing

  21. ALICE vs. LCG Service Challenges • We could: • Sample (bunches of) “RAW” events stored at T0 from our Catalogue • Reconstruct at T0 • Ship from T0 to T1’s • Reconstruct at T1 with calibration data • Store/Catalogue the output • As soon as T2’s start to join SC3: • Keep going with reconstruction steps + • Simulate events at T2’s • Ship from T2 to T1’s • Reconstruct at T1’s and store/catalogue the output

  22. ALICE vs. LCG Service Challenges • What would we need for SC3? • AliRoot deployed on LCG/SC3 sites - ALICE • Our AliEn server with: - ALICE • task queue for SC3 jobs • catalogue to sample existing MC events and mimic raw data generation from DAQ • UI(s) for submission to LCG/SC3 - LCG • WMS + CE/SE Services on SC3 - LCG • Appropriate amount of storage resources- LCG • Appropriate JDL files for the different tasks - ALICE • Access to the ALICE AliEn Data Catalogue from LCG

  23. ALICE vs. LCG Service Challenges • Last step: • Try the analysis of reconstructed data • That is SC4: • We have some more time to think about it

  24. ATLAS & SC3 • July: SC3 phase 1 • Infrastructure performance demonstration • Little direct ATLAS involvement • ATLAS observes performance of components and services to guide future adoption decisions • Input from ATLAS database group

  25. ATLAS & SC3 • September: SC3 phase 2 • Experiment testing of computing model • Running • 'real' software • production and data management systems • but working with throw-away data. • ATLAS production involvement • Release 11 scaling debugging • Debug scaling for distributed conditions data access, calibration/alignment, DDM, event data distribution and discovery • T0 exercise testing

  26. ATLAS & SC3 • Mid-October: SC3 phase 3 • Service phase: becomes a production facility • Production, adding more T2s • Operations at SC3 scale producing and distributing useful data • New DDM system deployed and operating • Conduct distributed calibration/align scaling test • Conduct T0 exercise • Progressively integrate new tier centers into DDM system. • After T0 exercise move to steady state operations for T0 processing and data handling workflows

  27. CMS & SC3 • Would like to use SC3 ‘service’ asap • Are already shipping data around & processing it as foreseen in SC3 service phase • More details in e.g. FNAL SC presentation • Next…

  28. CMS & Service Challenge III Service Challenge III - stated Goal by LCG: • Achieve 50% of the nominal data rate • Evaluating tape capacity to achieve challenge while still supporting the experiment • Add significant degrees of additional complexity • file catalogs • OSG currently has only experiment supported catalogs, CMS has a prototype data management system at the same time, so exactly what this means within the experiment needs to be understood • VO management software • Anxious to reconcile VO management infrastucture in time for the challenge. Within OSG we have started using some of the more advanced functionality of VOMS that we would clearly like to maintain. • Uses the CMS’s offline frameworks to generate the data and drive the data movement. • Requirements are still under negotiation

  29. LHCb • Would like to see robust infrastructure before getting involved • Timescale: October 2005 • Expecting LCG to provide somewhat more than other experiments • i.e. higher level functionality that file transfer service

  30. Experiment plans - Summary • SC3 phases • Setup and config - July + August • Experiment software with throwaway data - September • Service phase • ATLAS – Mid October • ALICE – July would be best… • LHCb – post-October • CMS – July (or sooner) • Tier-0 exercise • Distribution to Tier-1 • …

  31. SC3 – Milestone Decomposition • File transfer goals: • Build up disk – disk transfer speeds to 150MB/s • SC2 was 100MB/s – agreed by site • Include tape – transfer speeds of 60MB/s • Tier1 goals: • Bring in additional Tier1 sites wrt SC2 • PIC and Nordic most likely added later: SC4? • Tier2 goals: • Start to bring Tier2 sites into challenge • Agree services T2s offer / require • On-going plan (more later) to address this via GridPP, INFN etc. • Experiment goals: • Address main offline use cases except those related to analysis • i.e. real data flow out of T0-T1-T2; simulation in from T2-T1 • Service goals: • Include CPU (to generate files) and storage • Start to add additional components • Catalogs, VOs, experiment-specific solutions etc, 3D involvement, … • Choice of software components, validation, fallback, …

  32. A Simple T2 Model N.B. this may vary from region to region • Each T2 is configured to upload MC data to and download data via a given T1 • In case the T1 is logical unavailable, wait and retry • MC production might eventually stall • For data download, retrieve via alternate route / T1 • Which may well be at lower speed, but hopefully rare • Data residing at a T1 other than ‘preferred’ T1 is transparently delivered through appropriate network route • T1s are expected to have at least as good interconnectivity as to T0 • Each Tier-2 is associated with a Tier-1 who is responsible for getting them set up • Services at T2 are managed storage and reliable file transfer • DB component at T1; user agent also at T2 • 1GBit network connectivity – shared (less will suffice to start with, more maybe needed!)

  33. Prime Tier-2 sites • For SC3 we aim for • DESY   FZK (CMS + ATLAS) • Lancaster   RAL (ATLAS) • London   RAL (CMS) • Scotgrid   RAL (LHCb) • Torino   CNAF (ALICE) • US sites   FNAL (CMS) • Responsibility between T1 and T2 (+ experiments) • CERN’s role limited • Develop a manual “how to connect as a T2” • Provide relevant s/w + installation guides • Assist in workshops, training etc. • Other interested parties: Prague, Warsaw, Moscow, .. • Also attacking larger scale problem through national / regional bodies • GridPP, INFN, HEPiX, US-ATLAS, US-CMS

  34. SC3 – Milestone Decomposition • File transfer goals: • Build up disk – disk transfer speeds to 150MB/s • SC2 was 100MB/s – agreed by site • Include tape – transfer speeds of 60MB/s • Tier1 goals: • Bring in additional Tier1 sites wrt SC2 • PIC and Nordic most likely added later: SC4? • Tier2 goals: • Start to bring Tier2 sites into challenge • Agree services T2s offer / require • On-going plan (more later) to address this via GridPP, INFN etc. • Experiment goals: • Address main offline use cases except those related to analysis • i.e. real data flow out of T0-T1-T2; simulation in from T2-T1 • Service goals: • Include CPU (to generate files) and storage • Start to add additional components • Catalogs, VOs, experiment-specific solutions etc, 3D involvement, … • Choice of software components, validation, fallback, …

  35. Service Goals • Expect relatively modest increase in ‘service’ components • File catalog based on agreement from Baseline Services WG • Other services agreed by BSWG • Experiment-specific components and impact on other services, e.g. Distributed Database Services, need to be clarified as soon as possible • Similarly, requirements for processing power and storage at all sites involved (T0, T1, T2) • (This is for both Service and Challenge phases: where we run the experiments’ s/w and store the output!)

  36. SC3 – Milestone Decomposition • File transfer goals: • Build up disk – disk transfer speeds to 150MB/s • SC2 was 100MB/s – agreed by site • Include tape – transfer speeds of 60MB/s • Tier1 goals: • Bring in additional Tier1 sites wrt SC2 • PIC and Nordic most likely added later: SC4? • Tier2 goals: • Start to bring Tier2 sites into challenge • Agree services T2s offer / require • On-going plan (more later) to address this via GridPP, INFN etc. • Experiment goals: • Address main offline use cases except those related to analysis • i.e. real data flow out of T0-T1-T2; simulation in from T2-T1 • Service goals: • Include CPU (to generate files) and storage • Start to add additional components • Catalogs, VOs, experiment-specific solutions etc, 3D involvement, … • Choice of software components, validation, fallback, …

  37. Ramp-up of FTS and T1s • As expected, this is proving to be a significant amount of work! • Need to keep up momentum: • Increasing file transfer performance; • Adding tape; • Building up network links; • Bringing additional T1s into the challenges…

  38. File Transfer Milestones • Need to understand how infrastructure at CERN will build up to address file transfer goals of SC3 • Internal / External network infrastructure • File transfer Server nodes • Access to Experiment files • For each Tier1 need to understand how local infrastructure and network connectivity will be build up to meet requirements • For Tier2s, assume “Minimal Usable Model” (see earlier slide)… • Assume DB for FTS will run at T1 with “client” also at T2 • Transfers can be initiated from both ends

  39. 2005 2006 2007 2008 SC2 SC3 First physics cosmics SC4 First beams Full physics run 2005 Sep-Dec - SC4 preparation In parallel with the SC3 model validation period,in preparation for the first 2006 service challenge (SC4) – Using 500 MByte/s test facility • test PIC and Nordic T1s • and T2’s that are ready (Prague, LAL, UK, INFN, ..) Build up the production facility at CERN to 3.6 GBytes/s Expand the capability at all Tier-1s to full nominal data rate

  40. 2005 2006 2007 2008 SC2 SC3 SC4 First physics cosmics First beams Full physics run 2006 Jan-Aug - SC4 SC4 – full computing model services - Tier-0, ALL Tier-1s, all major Tier-2s operational at full target data rates (~2 GB/sec at Tier-0)- acquisition - reconstruction - recording – distribution, PLUS ESD skimming, servicing Tier-2s Goal – stable test service for one month – April 2006 100% Computing Model Validation Period(May-August 2006) Tier-0/1/2 full model test - All experiments - 100% nominal data rate, with processing load scaled to 2006 cpus

  41. 2005 2006 2007 2008 SC2 SC3 SC4 First physics cosmics LHC Service Operation First beams Full physics run 2006 Sep – LHC service available The SC4 service becomes the permanent LHC service – available for experiments’ testing, commissioning, processing of cosmic data, etc. All centres ramp-up to capacity needed at LHC startup • TWICE nominal performance • Milestone to demonstrate this 3 months before first physics data  April 2007

  42. June05 - Technical Design Report  Credibility Review by LHCC Sep05 - SC3 Service – 8-9 Tier-1s sustain - 1 Gbps at Tier-1s, 5 Gbps at CERN Extended peaks at 10 Gbps CERN and some Tier-1s Jan06 - SC4 Setup – AllTier-1s 10 Gbps at >5 Tier-1s, 35 Gbps at CERN July06 - LHC Service – All Tier-1s 10 Gbps at Tier-1s, 70 Gbps at CERN 2005 2006 2007 2008 SC2 SC3 SC4 First physics cosmics LHC Service Operation First beams Full physics run Key dates for Connectivity

  43. June05 - Technical Design Report Sep05 - SC3 Service Phase May06 –SC4 Service Phase Sep06 – Initial LHC Service in stable operation 2005 2006 2007 2008 SC2 SC3 SC4 First physics cosmics LHC Service Operation First beams Full physics run Key dates for Services

  44. Additional threads started to address: • Experiment involvement; • Bringing T2s in SC3; • Longer-term goals of bringing all T2s into the LCG Service (Challenges) • The enthusiasm and support provided to these new activities is much appreciated • We have a lot of work ahead… • …but the problem is beginning to become tractable(?)

  45. Conclusions • To be ready to fully exploit LHC, significant resources need to be allocated to a series of Service Challenges by all concerned parties • These challenges should be seen as an essential on-going and long-term commitment to achieving production LCG • The countdown has started – we are already in (pre-)production mode • Next stop: 2020

More Related