VDURA Data Platform adds snapshots, SMR support to boost scale and cut costs

Version 12 of VDURA Data Platform (VDP) scales out metadata compute, adds system-wide snapshots, and supports shingled magnetic recording (SMR) disk drives to lower per-TB costs.

VDURA says this release of its parallel file system software, known previously as PanFS, redefines scalability and resilience for AI and high-performance computing (HPC) environments, increases performance on every metric, and introduces a new Elastic Metadata Engine. This, plus the snapshots and SMR support, will “deliver unprecedented performance efficiency, durability, and capacity scaling across the data lifecycle.”

VDURA CEO Ken Claffey stated: “With VDURA V12, we’re increasing throughput performance by more than 20 percent, accelerating metadata operations by up to 20X, and reducing cost per terabyte by over 20 percent. At the same time, we’re simplifying data protection with new snapshot capabilities that make backup and recovery effortless at any scale.”

The Elastic Metadata Engine dynamically and linearly scales across metadata nodes to sustain performance as file and object counts grow into the billions, improving metadata operations by up to 20x. Snapshotting enables instantaneous, space-efficient point-in-time copies of datasets for AI pipelines, model checkpoints, and operational recovery. They are integrated with existing support for both manual and policy-based retention.

DDN’s ExaScaler Lustre-based systems support snapshots, as does IBM’s Spectrum Scale.

SMR disk drives are slower than conventionally recorded drives because they have zones of partially overlapping write tracks, increasing track density and hence capacity. However, whole zones have to be copied, edited, and rewritten if fresh data is to be added to any part of any track in the zone, slowing such rewrite operations.

VDP v12 has a new write-placement engine, which organizes sequential zones intelligently, unlocking, VDURA says, 25-30 percent more capacity per rack without compromising throughput or data integrity. This builds upon the data placement capabilities introduced last year in the v11 release.

It suggests that enterprises, AI service providers, and neoclouds will appreciate that Elastic Metadata Engine plus snapshotting plus SMR support reduces the system’s total cost of ownership and improves aggregate system throughput.

VDP v12 will be generally available in the second 2026 quarter for all V5000 and V7000-class systems. Existing customers running v11 or later can upgrade in place via an online software update with zero downtime.

Bootnote

Last year, at SC24, VDURA had Icelandic strong man Hafþór “Thor” Björnsson make a World Record Data Lift, hoisting over 282 PB of capacity, the most data ever lifted. This year, at SC25, VDURA Booth #2033, Björnsson will attempt to break the AI and HPC data transfer record: “Powered by Phison Pascari SSDs and the VDURA Data Platform, Thor will move hundreds of petabytes in seconds.” This year, VDURA says, it’s all about speed.