Service:storage: Difference between revisions

From OACISS Systems Wiki
Jump to navigation Jump to search
No edit summary
 
(19 intermediate revisions by 2 users not shown)
Line 1: Line 1:
= Storage on OACISS systems =
[[Category:Service]]
 
OACISS has multiple network attached storage systems for user operations. In addition, certain systems may have additional local disks, most often for storage of local images related to container infrastructure like Docker or Shifter.
 
Network attached storage operates over the 172.17 Ethernet, which is increasingly being upgraded to 100GbE performance.


[[Category:Service]]
== Main storage: /home/users ==
 
All systems in the CC utilize the main home directory NAS server on [[Infrastructure:mnemosyne | Mnemosyne]]. This is a primary storage engine with 12 15TB sas-3 SSDs in a raidz2 redundant storage array providing 128TB of disk in one volume.
 
Single-thread IO tests yield ~1.8GBps sequential read and ~10K IOPS 4K random r/w.
 
== Package tree: /packages ==


OACISS runs two NAS storage drives for user data and programs.
Package trees for RH7, RH8 and ppc64le are stored on 15TB SAS-12G SSDs in RAID-1.


In addition, certain systems may have additional local disks, most often for storage of images related to container infrastructure like Docker.
As packages are subject to almost exclusively read access, they are mounted synchronously.


== Main storage ==
== High performance storage: /gpfs/gpfs0 ==


Most systems access NAS on [[Infrastructure:mnemosyne | Mnemosyne]]. This is the primary storage engine utilizing 14x12TB Seagate Exos disks in RAID-6. It has 135TB of disk available in a single large volume.
OACISS has deployed an IBM GS4S high-performance file server utilizing the Spectrum Scale storage engine. The ESS server cluster has 96 4TB SSDs providing a total of 250TB of storage. The ESS cluster is linked to the private storage network with 200Gbps of Ethernet bandwidth.


Mnemosyne is connected to the storage network at 40Gbps and has 128GB of memory. It can absorb data at 2.5GBps until ram cache is filled, and can sustain roughly 750MBps sequential IO continuously.
The underlying high-performance Spectrum Scale filesystem driver is able to deliver up to approximately 3GBps of sequential IO per thread.


Package trees for architectures and various major versions of OS are exported and normally mounted sync. All user directories are mounted NFS4 async.
== BeeGFS storage ==


== NVME NFS ==
The majority of nodes have a high performance BeeGFS filesystem available.


In addition, Mnemosyne hosts and exports an 8TB volume made from two 4T Sabrent Rocket NVME gumsticks. This volume is capable of delivering 2.5GBps of read and write simultaneously nonstop (limited to somewhat under 1GBps per client by the client network speed), or up to 80K IOPS, over the storage network.
The data storage lives on a dedicated host, Alexandria, with 15 x 15TB NVME drives arranged in a 180TB raidz2 pool.


== Cerberus ==
== Upstairs storage ==


Cerberus, Chymera, and the nuc + jetson cluster have access to a smaller 22TB NAS, accessed over 1Gbit ethernet, located on Cerberus.
Upstairs storage is a 28TB SSD storage pool living on Sphinx, serving Silicon and the ARM hardware cluster (jetsons, xaviers, orins and OD1K).


This storage, and package tree, is entirely separate from main storage.
This storage, and its package tree, are entirely separate from main storage.

Latest revision as of 01:34, 30 November 2023


OACISS has multiple network attached storage systems for user operations. In addition, certain systems may have additional local disks, most often for storage of local images related to container infrastructure like Docker or Shifter.

Network attached storage operates over the 172.17 Ethernet, which is increasingly being upgraded to 100GbE performance.

Main storage: /home/users

All systems in the CC utilize the main home directory NAS server on Mnemosyne. This is a primary storage engine with 12 15TB sas-3 SSDs in a raidz2 redundant storage array providing 128TB of disk in one volume.

Single-thread IO tests yield ~1.8GBps sequential read and ~10K IOPS 4K random r/w.

Package tree: /packages

Package trees for RH7, RH8 and ppc64le are stored on 15TB SAS-12G SSDs in RAID-1.

As packages are subject to almost exclusively read access, they are mounted synchronously.

High performance storage: /gpfs/gpfs0

OACISS has deployed an IBM GS4S high-performance file server utilizing the Spectrum Scale storage engine. The ESS server cluster has 96 4TB SSDs providing a total of 250TB of storage. The ESS cluster is linked to the private storage network with 200Gbps of Ethernet bandwidth.

The underlying high-performance Spectrum Scale filesystem driver is able to deliver up to approximately 3GBps of sequential IO per thread.

BeeGFS storage

The majority of nodes have a high performance BeeGFS filesystem available.

The data storage lives on a dedicated host, Alexandria, with 15 x 15TB NVME drives arranged in a 180TB raidz2 pool.

Upstairs storage

Upstairs storage is a 28TB SSD storage pool living on Sphinx, serving Silicon and the ARM hardware cluster (jetsons, xaviers, orins and OD1K).

This storage, and its package tree, are entirely separate from main storage.