Software-defined Storage Architecture for Analyltics Computing

Software Defined Storage is modifying the way storage is being consumed. With the increase of cores per server and server-side storage slots available, software is the key aspect to virtualize all those components and unlock the performance and capacities of faster networks and processors.

While VERITAS has being virtualizing storage for decades, its new generation software allows cost-effective solutions where SAN is no longer needed for traditional workloads avoiding high investments that were needed before. The architecture proposed in this white paper is able to read data at 15GB/s just using commodity servers and HDDs. Think in a traditional SAN how many HBA’s are needed to provide a similar performance.

Software really unlocks the potential of HW. And remember, a truly SDS solution should be agnostic so it should avoid any HW lock-in.

This White Paper presents a typical architecture for a leading analytics application like SAS Grid, where traditionally a SAN is needed. Using severs with 25 internal HDDs we can achieve a traditional SAS Grid architecture that satisfices their high I/O requirements.

analytics.jpg