NSULATE is a next-generation alternative to RAID for modern big data and HPC applications

Say goodbye to RAID

RAID is obsolete and an obstacle to achieving hyperscale storage.

RAID was standardized in 1993, before the Internet age, in an era of single-core computing and spinning magnetic hard disks. NSULATE substitutes modern highly parallel general purpose GPU’s for traditional ASIC-based RAID functions, enabling massive real-time acceleration of high parity erasure encoding, data compression, deduplication, encryption, caching, storage tiering and data analytics in a low-level block device compatible with all existing Linux filesystems, storage solutions and storage devices.

Hyperscale Erasure Coding

NSULATE scales erasure encoding calculations by two orders of magnitude in real-time, supporting up to 256 parity in a single contiguous array. Higher parity means that more physical storage devices can be used in parallel, enabling higher storage performance, resilience, security and power efficiency with fewer redundant storage resources than any other known storage technology available today.

Unlimited Scalability

NSULATE enables distributed storage arrays that self-tier and scale dynamically as more storage devices are added to the array. Device addressing in an NSULATE array is entirely virtualized so that arrays can appear to have unlimited storage and can be over provisioned as needed. More storage nodes, storage devices and tiers can be freely added or removed from a live functioning array and NSULATE will dynamically adapt and optimize as needed.

Storage Processing


The NSULATE block device is powered by a GPU-accelerated graph engine (similar to Apache Spark). Features like encryption, compression and real-time data analytics can be added to the storage graph and applied in-line as part of data storage transactions. NSULATE turns a traditional static storage pipeline into a general purpose stream processing architecture that can automatically perform general purpose analytics functions on incoming or outgoing data streams.