VAST Simulation Finds Higher Data-Loss Odds for Petabyte SSDs
At Supercomputing 2026, VAST Data presented a discrete-event simulation of its 900 PB storage system. The model estimated a 1-in-11 chance of data loss over 10 years when petabyte-scale SSDs are used with enclosure-local declustered RAID, according to the company. VAST recommended system-wide erasure coding with at least three parity strips; the report says global declustering with extra parity could reduce the modeled odds to 1 in 54 million, while VAST’s own approach reached about 1 in 62 billion. The company also modeled 900 PB of 15 TB flash requiring more than 63,000 drives, 190 TB/s of bandwidth and 5.6 MW of power. By comparison, its estimate for nearly 8,000 122 TB drives was 8 TB/s and about 400 kW. The report says open, fair comparison is still needed before drawing a firm conclusion about the best approach.
Larger drives take longer to rebuild after failures, increasing the window for additional failures to cause data loss. The simulation makes reliability, rebuild performance, power and bandwidth central trade-offs for AI and HPC storage design, while the need for open comparison limits any firm conclusion about the preferred protection scheme.
VAST Data says petabyte SSDs will outgrow RAID in today’s parallel file systems, according to a simulation of its own 900 PB system that puts the odds of data loss at 1 in 11 over 10 years. At Supercomputing 2026 (SC26), the company explains how.
Bigger drives take longer to rebuild after a failure, leaving more time for further failures to lose data, but operators will still want those large drives because many sites are power-constrained and storage power scales with bandwidth rather than capacity. VAST’s recommended fix: system-wide erasure coding with at least three parity strips.
The company’s principal technical strategist, Glenn Lockwood, walked through the simulation, described as a “discrete event simulation” using real failure data, in an Exhibitor Forum talk during SC26’s HPC System Hardware and Architecture session. With petabyte-scale SSDs swapped in, enclosure-local declustered RAID — the approach used by DDN EXAScaler and HPE Cray ClusterStor, per Lockwood — has those exceptionally high odds of data loss. With global declustering and extra parity, operators can get this down to 1 in 54 million, a significant improvement, but VAST’s own scheme gets to about 1 in 62 billion, three orders of magnitude better.
#VAST Data#DDN EXAScaler#HPE Cray ClusterStor#Lustre#BeeGFS#Micron 6600 ION#Kioxia LC9#SanDisk UltraQLC#Solidigm