Kaminario announced a new video, Kaminario's Scale-out Storage Performance Architecture Revealed, featuring Vice President of Engineering Shachar Fienblit. In the video, Fienblit discusses the Kaminario K2 all solid-state SAN storage architecture, a unique architecture built specifically for high performance SDD media.

Legacy architecture was built for slow, mechanical hard disk drives (HDDs), which Kaminario realized would not work for fast SSD media because there are so many bottlenecks in the system that would prevent utilizing SSD media to accelerate application and database performance. The Kaminario architecture, called SPEAR (Scale-out Storage Performance Architecture), was designed with three goals in mind. Fienblit details these goals:

"The most important goal was performance. We built the architecture to deliver the highest performance, the lowest latency, and the highest IOPS and throughput. Along with performance, we realized high availability is important for mission-critical environments; this is why we built a system without any single point of failure and with all the data fully available and all the recovery operations occurring automatically. The third goal was that the system should be easy, which means no tuning, as opposed to a legacy system where tuning is required and a large number of operations are required to configure the system."

To achieve those goals, Fienblit explains:

"We built a system from best-of-breed X86 servers, with best-of-breed Flash and DRAM SSD media. The architecture features the I/O directors we use for performance, the data nodes that are required for capacity and are able to scale independently, and the SPEAR operating system, which is managing all the features together, delivering a system offering high availability and data protection and managing the entire Kaminario scale-out shared storage cluster. What we have built is a state-of-the-art, enterprise-grade, high performance storage system done correctly."

Fienblit has an extensive background in storage that includes 10 years with IBM as its lead storage architect for functions such as snapshots, replication, thin provisioning and other disaster recovery solutions.