COMPUTE ATLAS Supercomputer supply-chain graph
1153 systems 482 sites Sourced data

Parts/Storage/Distributed RAID 5 disk array

Storage · 1996

Distributed RAID 5 disk array

Direct-attached disks striped in RAID 5 across I/O units, the storage architecture that predates any parallel filesystem in this dataset.

Why this part

What every large machine used before a parallel filesystem existed to buy. CP-PACS held 529 GB across its I/O units in a RAID 5 configuration, no Lustre, no GPFS, both a decade away from existing as products. RAID 5 stripes data across disks with a rotating parity block, so a single drive failure costs no data, which was the entire point once a machine had enough disks that one failing during a run stopped being a rare event and started being a certainty. Every dedicated parallel filesystem part in this taxonomy exists to solve the problem this approach could not: letting thousands of compute nodes write to the same storage at once rather than one I/O node serving them in turn.

Systems using this part

Systems
1
Rmax underneath
368.2 GFlop/s
Units recorded
-
0 of 1 edges disclose a quantity
System Role Country Units System Rmax Supplied by Rank
CP-PACS
1996-10
Storage Japan - 368.2 GFlop/s Hitachi #1 →

“Units recorded” sums only the edges where a public source states a quantity. It is a floor, not a total, and should never be read as installed base.

Further reading