Service:storage: Difference between revisions

From OACISS Systems Wiki
Jump to navigation Jump to search
No edit summary
Line 9: Line 9:
= Main storage =
= Main storage =


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.
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. Vital statistics:


Mnemosyne is connected to the storage network at 40Gbps and has 128GB of memory. It can absorb data at 2.5GBps until RAM runs out, and can sustain roughly 750MBps sequential IO continuously. Any single node is limited to roughly 1GBps of IO by its own 10Gbit link.
* 40Gbps Ethernet link
 
* 128G RAM
For performance reasons, main storage is mounted with -oasync.
* ~700MBps sustained sequential write
* ~3.5GBps sustained sequential read
* Mounted with -oasync for performance


= Package tree =
= Package tree =

Revision as of 08:46, 20 November 2020

Storage on OACISS systems

OACISS has three separate NAS storage drives, for bulk user data, programs and scratch.

In addition, certain systems may have additional local disks, most often for storage of images related to container infrastructure like Docker.

Main storage

Most systems access NAS on 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. Vital statistics:

  • 40Gbps Ethernet link
  • 128G RAM
  • ~700MBps sustained sequential write
  • ~3.5GBps sustained sequential read
  • Mounted with -oasync for performance

Package tree

Package trees for RH7, RH8 and ppc64le are stored on two 4TB SAS-12G SSDs in RAID-1.

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

High performance scratch

Mnemosyne hosts a high-performance 8TB scratch volume made from two 4TB Sabrent Rocket NVME M.2 SSDs.

This volume is available on the 10Gbit network and on the 100Gbit InfiniBand network; Any hosts which have IB access mount it over the IB network.

The scratch volume is capable of delivering 1.1GBps and 80K IOPS per host over the 10G network (to a limit of about 2.5GBps and 95K IOps total). Over the 100G links, it is capable of delivering an astounding 10GBps and 550K IOPS (sustained sequential write limit ~7GBps).

Upstairs storage

Cerberus, Chymera, and the nuc + jetson cluster have access to a smaller 22TB NAS, accessed over 1Gbit ethernet, located on Cerberus.

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