Scality just rolled out Ring XP, an enhanced version of its Ring platform, specifically designed for AI data lakes. This upgrade promises super-fast storage access times, clocking in at just microseconds.
What’s driving this upgrade? A simplified version of S3 object storage APIs that cut out unnecessary features that slow things down. They’ve paired this with all-NVMe flash hardware. Paul Speciale, Scality’s chief marketing officer, explains that customers are keen on aggregating massive amounts of data while needing rapid input/output for tasks like data cleansing, security, immutable backups, and heavy metadata operations. He highlights that Ring XP is tailor-made for the small object data typical in AI processing.
So, what’s small object data? Think tiny fragments from huge datasets—like pieces of images and short strings in log files. Scality believes Ring XP can stand toe-to-toe with high-performance storage options from companies like IBM, Pure Storage, and Vast Data.
To deliver on that performance, Ring XP uses a simplified S3 connector and relies on all-NVMe flash storage. It contrasts sharply with the traditional S3 API options, which come loaded with features like versioning and replication but have access times between 10 milliseconds and 50 milliseconds.
AWS does have options like the S3 Express Zone One, which strips away some features to achieve 1 to 9 milliseconds of latency. Ring XP, however, goes even further, promising less than 500 microseconds for read operations and under 700 microseconds for writes. Speciale points out that while Amazon’s S3 has an extensive feature set, it brings significant latency. He mentions that Amazon has recognized this and provides faster options, but Scality claims to push the boundaries further. They assert that no other storage provider has published latency figures that low.
For existing Ring customers, switching to Ring XP is straightforward—just pay the licensing fee and make sure the right hardware is in place.
Scality has two key products: Artesca and Ring, targeting different market segments. Ring is tailored for enterprise-level storage at a minimum of 5PB, while Artesca serves smaller businesses that want object storage solutions with on-premise applications.
As for real-world applications, a US bank uses Ring to aggregate its fraud detection data via Splunk, managing tens of petabytes of information. Meanwhile, Amadeus, a European travel pricing platform, handles about 1PB of data daily through Scality for its Splunk-based analyses. On the genomics side, French firm SeqOIA has integrated Ring as its bulk storage. They even leverage its speed for demanding tasks. They do have a Vast Data setup for ultra-high-performance needs, but as Speciale indicates, they haven’t felt the need to expand that given Ring’s capabilities in managing large-scale data and diverse operational tasks.