Infrastructure of the Tier1 (T1_RU_JINR)

The service support system provides the following:

  • computing service,
  • storage services,
  • GRID systems,
  • data transfer services ,
  • information service (monitoring services:  servers, storage, data transfer, information sites).

The Slurm workload manager is used to launch data processing tasks. The grid environment uses Advanced Resource Connector (ARC), a grid computing middleware. It provides a common interface for transferring computing tasks to various distributed computing systems and can include grid infrastructures of varying size and complexity.

The dCache and EOS systems are used as the main data storage systems.  The afs is used  for user home directories,  cvmfs storage systems is  used to store project software (cvmfs is a system  for distributed access and organization of software versions for collaborations and user groups, cvmfs is used to store project software).

Tier1 farm configuration (T1_RU_JINR):

The CMS experiment data processing jobs are launched by 16 core pilots and all computing resources are available to them.

Computing farm (CE)

Sum 2023Q4:
323820.54 HEP-SPEC06 
20096 cores 
468 Wns
Average per Core = 16.11 

Storage Systems (SE)

Data storage is provided for the dCache system, robotic tape storage using Enstore software and the EOS system.

2023:

dCache:  SE disks:
        2023Q1=11.02 PB;  
        2023Q2=11.76 PB; 
        2023Q4=12.43987 PB;         
2023Q4: 
         CMS @ dsk:       12439.87 TB 
         dteam @ dsk:         3.21 TB 
         OPS @ dcache dsk:    4.36 TB
         CMS @ dcache mss: 2657.20 TB  
         DTEAM @ dcache mss   4.51 TB
         OPS @ dcache mss     4.51 TB
         Tapes@Enstore:    35562.00 TB 
                      
tape: allocated capacity for T1 logic library 51.5PB :  
      IBM TS3500(11.5PB for EOSCTA) + 
      IBM T4500(40PB=20PB for CMS + 20PB for NICA )
 
EOSCTA: 11500 TB;

         
EOS:    2023Q4=23328.10 TB;
        2023Q3=22203 TB;   
        2023Q2=21829.01 TB;   
        2023Q1=16582.35 TB; 


CVMFS:   capacity is 140TB
         4 squid servers cache CVMFS 
         1 server stratum-0,
         2 servers stratum-1.

Total: EOS(23328.10 TB) + dcache_disk(12439.87 TB) + tapes(51500 TB) + CVMFS(140 TB) = 87407,97 TB


2023Q2: dCache – added 6 servers (Qtech QSRV-462402_4) = ~2.3 PB ; decommissioned – 15 servers (Supermicro rda017-rda031~2017) = ~1.719 PB

 

Software :

2023Q1

ОС: Scientific Linux release 7.9.

EOS 5.1.23 (constantly changing),

dCache 8.2, 

Enstore 6.3.

Slurm 20.11.

grid UMD4 + EPEL (current versions)

Batch Slurm 20.11

grid UMD4 + EPEL, актуальные версии
ARC-CE


 

 

 


Infrastructure of the Tier1 (T1_RU_JINR) 2023

Infrastructure of the Tier1 (T1_RU_JINR) 2022

Infrastructure of the Tier1 (T1_RU_JINR) 2021

Infrastructure of the Tier1 (T1_RU_JINR) 2020

Infrastructure of the Tier1 (T1_RU_JINR) 2019

Infrastructure of the Tier1 (T1_RU_JINR) 2018

Infrastructure of the Tier1 (T1_RU_JINR) 2017

Infrastructure of the Tier1 (T1_RU_JINR) 2016