HPE Introduces Next Generation Storage Class Memory Technology Across Its Storage Systems Portfolio
DECEMBER 13, 2018 • BLOG POST • ERIC BURGENER, RESEARCH VICE PRESIDENT, IDC
IN THIS ARTICLE
- SCM technology is enabling new, high performance storage to address the increasingly stringent performance requirements of many new workloads enterprises need to support as part of digital transformation
- HPE worked closely with Intel to create HPE Memory-Driven Flash, an optimized, enterprise shared storage solution
IDC's Eric Burgener dives into the next generation of storage - storage class memory (SCM) and HPE's Memory-Driven Flash
You may have heard about storage class memory (SCM), a new persistent tier in the memory/storage hierarchy that sits (roughly) between DRAM and NAND flash in terms of performance, capacity and cost. Evolving storage workloads, including next generation applications that leverage artificial intelligence, machine learning, and real-time analytics, are driving the need for a performance tier beyond what today's all flash arrays can offer.Evolving storage workloads, including next generation applications that leverage artificial intelligence, machine learning, and real-time analytics, are driving the need for a performance tier beyond what today's all flash arrays can offer.SCM technology will enable a new, high performance persistent storage tier that is needed to address the increasingly stringent performance requirements of many new workloads enterprises need to support as part of digital transformation. HPE has now partnered with Intel to bring SCM technology to their proven enterprise storage platforms, HPE 3PAR and HPE Nimble Storage.HPE has now partnered with Intel to bring SCM technology to their proven enterprise storage platforms, HPE 3PAR and HPE Nimble Storage.
Which Workloads Benefit From SCM?
Workloads where lower latencies drive improved business results are driving many early SCM deployments. Transactional databases that drive revenues in industries like financial services and e-commerce are a prime candidate for SCM, with stock exchanges being among the early adopters. Real-time analytics workloads driving time-sensitive results for on-line ad placement, customized offer presentations to in-store shoppers, and promotions tied to discrete events (e.g. a World Cup soccer match, an Olympic swim meet, etc.) that key off of social media profiles with location dependencies are other use cases where IDC has seen SCM deployed to drive faster results. Artificial intelligence and machine learning-based inference engines that must comb through massive constantly changing data sets to drive real-time recommendations are another key area, and the large growth of these types of workloads across virtually all industries going forward will drive SCM revenues. The sheer performance of this new class of storage will enable many types of workloads that were not even feasible just two years ago.
Consolidate and Simplify Infrastructure with Predictable and Consistent Performance at Scale
As part of infrastructure modernization efforts, many customers are looking to make better use of their IT budget. One of the strategies CIOs often use is the consolidation of storage workloads onto fewer platforms. Enterprise-class storage platforms offer a full suite of storage management features that help to make denser consolidation easier, but there are performance concerns as well: IT managers must continue to be able to meet service-level agreements they've made to their customers even as they increase workload density on a single platform. Denser consolidation, when feasible, results in fewer systems to manage and maintain, reduces administrative complexity, uses datacenter floor space more efficiently (making more room for growth), and generally improves TCO. With its high-performance characteristics, SCM delivers more predictably consistent response times at higher workload densities even as it lowers latencies. As enterprises look to further consolidate workloads, SCM will play an important enabling role.
HPE an Early Entrant in the Memory-Driven Storage Space
The coupling of SCM with proven enterprise storage platforms like HPE 3PAR and HPE Nimble Storage will be the preferred deployment model for this use case. HPE's recent HPE Memory-Driven Flash announcement allows customers to non-disruptively upgrade existing arrays with an Intel Optane solid-state disk-based caching tier, providing the performance needed for the most demanding workloads while at the same time enabling denser workload consolidation. HPE worked closely with Intel to create an optimized, enterprise shared storage solution for HPE Memory-Driven Flash:
"Since previewing SCM in 2016, HPE has collaborated with Intel to make Intel Optane technology ready for the enterprise," said Bill Leszinske, Corporate Vice-President of Intel's Non-Volatile Memory Group. "I'm excited that HPE is delivering this new enterprise storage solution. HPE Storage, with Intel Optane technology, is leading a fundamental shift in storage architecture that will drive the next generation of innovation."
As they are among the first-movers to combine SCM performance with true enterprise-class capabilities, they are instrumental in driving this market forward and opening up additional opportunities for operational efficiencies with denser workload consolidation.
What Enterprises Most Want from SCM Are Predictably Consistent Low Latencies
This past summer, IDC did primary research that explored customer perceptions of and experience with SCM, generating some interesting findings. SCM requires NVMe, a new storage protocol that leverages the performance and capacity of memory-based media like flash and SCM much better than SCSI (a protocol which was developed specifically for hard disk drives). Awareness of SCM is very high among NVMe users, and 95.0% of the NVMe users surveyed worldwide were somewhat or very interested in testing SCM. Performance is clearly the driving factor, with customers looking to it to provide more predictably consistent low latencies under load as well as higher throughput and bandwidth. Survey respondents confirmed that the workloads they will be targeting for SCM deployment include in-memory databases, real-time analytics, and more technically-oriented computing for business (e.g. genomics in healthcare, seismic processing in oil and gas, modeling and simulation for design purposes in various industries, etc.). The survey confirmed that it is early in the growth of the SCM market, but that it will evolve quickly over the next two years. 15.7% of survey respondents expected to be testing SCM by the end of 2018, and 62.1% expect to be testing it by the end of 2019.
HPE is Delivering the SCM Performance Boost Today
Customers needing more storage performance for enterprise-class workloads can look to SCM now for performance boosts. Vendors like HPE whose deployment strategy makes it easy to turbo-charge existing storage with this technology offer the best opportunity to bring SCM on board with the least disruption.