Flash Memory Gives Databases a Jolt
Database vendors are looking for new memory technology options as datasets continue to soar. With that in mind, Redis Labs Inc. claimed this week that its database running on flash memory along with Intel Corp.’s NVM Express flash-based SSDs achieved record performance.
Redis Labs, Mountain View, Calif., announced Thursday (Aug. 11) during the AWS Summit in New York City that the combination of flash memory and Intel’s NVMe flash drives yielded a benchmarked throughput of 3 million database operations per second at under 1 millisecond of latency. The configuration generated more than 1 Gb of NVMe throughput on a single x86 server running on the Redis database, the partners claimed.
Benchmark reports for the flash-based Redis database running on both Intel’s (NASDAQ: INTC) and Samsung’s NVMe drives are available here.
Redis Labs said its database leverages flash memory as a “RAM extender,” noting the flash memory delivers “memory-like performance” at a cost estimated to be as low as one-tenth the cost of DRAM.
As database deployments expand both in size and in the variety of use cases, Redis Labs argued that the challenges posed by rising memory costs could be overcome with advanced memory technologies like Intel’s NVMe flash.
Redis running on flash is available now in Redis Labs Enterprise Cluster and the company’s Cloud Private Flash platform. The cloud service is designed to run Redis datasets on Amazon Web Services’ (NASDAQ: AMZN) virtual private cloud. The database operates on dedicated clusters, and the company is offering the option of allocating “use-case appropriate” datasets between RAM and flash memory using high IOPS SSD instances.
For its part, the chipmaker claims its next-generation memory technology would deliver substantial improvements in latency “so that data tiers such as what Redis Labs has developed will continue to blur the lines of what non-volatile memory can do for efficient caching and data persistence strategies,” noted Frank Ober, Intel’s datacenter solution architect.
Along with its collaboration with Intel, Redis Labs announced in April it would partner with South Korean memory chip powerhouse Samsung Electronics (KRX: 005930) to accelerate the processing and analysis of bulging datasets using next-generation memory technology designed to significantly cut memory costs.
Redis also joined forces with Samsung to demonstrate 2 million operations per second throughput with sub-millisecond latency along with 1 Gb of disk bandwidth running on a standard Dell Xeon-based server. Eighty percent of the dataset was processed using the NVMe SSD technology while the remaining 20 percent was off-loaded to standard DRAMs.
The combination of Samsung’s next-generation NVMe-based drives and Redis flash-based processing seeks to leverage HPC performance for crunching extremely large datasets. In benchmark tests, the partners claimed the Samsung drives delivered a 40-fold increase in throughput. At the same time, they said the cost of the new SSDs “is likely to be only incremental compared to [Serial Advanced Technology Attachment]-based SSDs.”