Accesso Utenti

Computation Data Analysis

Computation Infrastucture

It is important to optimize the scientific impact for the INAF community by creating optimal conditions for data access and analysis. The analysis of LOFAR data requires very complex computational procedures and infrastructures suited to the size of the datasets produced by the individual observations, typically 10-20 TB.
Upon entering in LOFAR, INAF had no direct/reserved access to these infrastructures.
For this reason we have planned investments to provide about 20 users, distributed at italian national level, with proximity calculation resources to conveniently analyze the LOFAR data.
Considering the involvement of the University of Turin, the competences distributed in INAF and the resources already available, LOFAR IT will have a calculation structure organized in 2 levels (Platform 1 and 2):

A Platform1 consisting of a single system with 10GB connection.
Considering UniTO's willingness to make 3 FAT nodes available in OCCAM and 150 TB storage at full speed and 10GB connection (pdf), this system has been identified as adequate for Platform1 (Platform1-TO), as already foreseen by the agreements signed [attachment 4]. At the moment the Platform1 consists of FAT nodes as follows:
- CPU - 4x Intel® Xeon® Processor E7-4830 v3 12 core/2.1Ghz
- RAM - 768GB/1666MHz (48 x 16Gb) DDR4
- DISK - 1 SSD 800GB + 1 HDD 2TB 7200rpm
- NET - IB 56Gb + 2x10Gb
and 50 TB initially, with the possibility to upgrade since to 150 TB over 18 months

A Platform2 consisting of a network of systems (proximity calculation) with 3 geographical nodes that have similar
characteristics and that share a common infrastructure (e.g. common authentication system, filesystem possibly shared,
common system softare).
The nodes that will host the Platform2 have been identified in IRA, OA Catania and OA Trieste.
On the basis of the tests developed during 2018 for each site of the Platform2, 3-5 nodes composed of 2 sockets with 40 cores (Intel E5
or later type) with at least 256 GB RAM are suggested.
At the moment Platform2 consists of:

Platform2 IRA :

RAM CPU Cores Clock Data Net Work Disk
(DAS HD)
Scrach Disk
(DAS SSD)
512G Intel Xeon E5-2640 v4 2 x 10/20 2400/3400 1GbE 28TB 196GB
384G Intel Xeon Gold 6130 2 x 16/32 2100/3700 10GbE 19TB 65GB
384G Intel Xeon Gold 6130 2 x 16/32 2100/3700 10GbE 10TB 65GB

Currently used by (oppure reserved to) DRANOEL group

RAM CPU Cores Clock Data Net Work Disk
(DAS HD)
Scrach Disk
(DAS SSD)
512G Intel Xeon E5-2640 v4 2 x 10/20 2400/3400 1GbE 28TB 196GB
384G Intel Xeon Gold 6130 2 x 16/32 2100/3700 10GbE 19TB 65GB
384G Intel Xeon Gold 6130 2 x 16/32 2100/3700 10GbE 10TB 65GB
384G Intel Xeon Gold 6130 2 x 16/32 2100/3700 10GbE 10TB 65GB
384G Intel Xeon Gold 6130 2 x 16/32 2100/3700 10GbE 10TB 65GB

Storage
240 TB RAID storage server


Platform2 OA CT :

RAM CPU Cores Clock Data Net Work Disk
(DAS HD)
256G Intel® Xeon® E5-4627 4 x 10 2600/3600 10GbE 30TB
256G Intel® Xeon® E5-4627 4 x 10 2600/3600 10GbE 30TB
512G Intel(R) Xeon(R) Gold 5118 4 x 12 2300/3600 10GbE 30TB

Storage
70 TB storage server


Platform2 OA TS :

RAM CPU Cores Clock Data Net Work Disk
(DAS HD)
Scrach Disk
(DAS SSD)
512G Intel(R) Xeon(R) Gold 5118 4 x 12 2300/3600 Infiniband HDR56  80TB 128GB
512G Intel(R) Xeon(R) Gold 5118 4 x 12 2300/3600 Infiniband HDR56 80TB 128GB
512G Intel(R) Xeon(R) Gold 5118 4 x 12 2300/3600 Infiniband HDR56 80TB 128GB
512G Intel(R) Xeon(R) Gold 5118 4 x 12 2300/3600 Infiniband HDR56 80TB 128GB

Storage
500TB beegfs parallel storage

Data Analysis Software

The Italian LOFAR group provides containerized versions of LOFAR pipeline software to provide:

  • software portability
  • correctness of the software version management
  • result reproducibility

The containerized software with the documentation is available at https://www.ict.inaf.it/gitlab/lofarit/containers.