Meeting with IT (Storage for O2)

Europe/Zurich
6th of January, 2021
 
Present: L. Betev, F. Hemmer, M. Lamanna, L. Musa, A. Morsch, A. Pace, B. Panzer, S. Piano, M. Ploskon, W. Riegler, W. Salter
 
  • ALICE presented estimates of the data rate and size (SP slides)
    • Given 2.7/nb integrated luminosity in 2022 PbPb collisions ALICE requires 80PB of raw capacity of the disk buffer.
    • Additional considerations of contingency related to the LHC performance, detector performance, compression efficiency amount to 20%. In general, ALICE is interested in estimates of IT's capacity for provision of operational / ad-hoc contingency helping accommodating for, for example, an increased LHC performance.
    • Full-scale storage tests (including connectivity/throughput) to be performed will be detailed in upcoming weeks (ML) - see below
  • The disk to cover ALICE needs has been already ordered and is to be installed in the CC (adding to the existing 20PB). From the 80PB a fraction (up to 20) is allocated to ad-hoc non-ALICE testing throughout 2021. ALICE requires the disk buffer operational in 2021 to complete full-scale commissioning.
  • Depending on the outcome from the foreseen tests the discussion on finances will take place later in the summer of 2021 . Concurrently, in the upcoming months ALICE funding for the disk buffer will be consolidated vis-à-vis other computing project expenditures.
  • One of the critical items brought up was the replica strategy - currently the target is a 10+2 configuration (initially evaluated possible). The change to a less aggressive scenario (e.g. 8+4) would be equivalent to a rise of the effective storage cost. Clearly, the decrease in the useable disk space may compromise the recorded luminosity thus impact ALICE's physics programme.
    • It was remarked that running the 10+2 configurations might require a production agreement between O2 and IT ST detailing the specifications of the expected performance of the system (e.g. short-term unavailability of data subset during rebuilds).
  • P2-CC connectivity: the 1 Tb is in place as it will be tested as soon the ALICE IB-ENET gateway arrives (expected on Jan 11). The 2 Tb is agreed. The possibility for a redundant physical path to CC is desirable and should be investigated by IT (this entails a second set of 1Tb links going to Meyrin through the Prevessin site complementing the existing path).
  • ALICE will prepare a detailed list of tests to be done (by ALICE + IT depending of the type of tests). This is linked with the commissioning test of ALICE (the so-called Milestone Weeks - monthly integration exercises).
  • FH informed ALICE about the new computer center (PCC as Prevessin Comp. Centre to be built close to the CCC). Building capable of 12MW (civil engineering) ready to deliver 4MW ready in 2023. 8MW and 12MW pre-approved by FC but needs and budget to be analysed.
  • ALICE informed IT about the availability of some capacity in the P2 containers (~1MW). BP answered this could be part of some 'plan-b' scenarios.

 

There are minutes attached to this event. Show them.
    • 09:00 09:10
      Introduction 10m
    • 09:10 09:25
      Status of O2 data rates and sizes 15m
      Speaker: Stefano Piano (INFN (IT))
    • 09:25 10:00
      Discussion 35m