Advanced Computing in the Age of AI | Thursday, April 18, 2024

New All-Flash from Nimbus, Storage-Class Memory from Netlist 

Nimbus Data and Netlist unveiled new storage and memory products of major potential at the Flash Memory Summit in Santa Clara this week.

Nimbus launched an all-flash platform for cloud and big data, called ExaFlash, that company CEO and founder Thomas Isakovich said “will reshape the storage and data center industries,” with pricing at less than half ($0.19 per effective gigabyte) of existing all-flash systems, 95 percent lower power consumption, 50X greater density and 90 percent lower total cost of ownership. He also said the platform scales up to exabytes.

“ExaFlash brings the all-flash data center dream to reality,” Isakovich said in a webinar introducing the product. Exaflash offers a new all-flash storage architecture that builds on, and overcomes the limitations of, scale-up and scale-out storage architectures, which Isakovich said were built for disk. The ExaFlash architecture enables “virtually infinite scalability,” Isakovich said, as one centrally-managed all-flash system. Data flow is decoupled from metadata and management is centralized, operating completely out-of- band, providing one interface for all administration.

“There’s no network between the Exaflash arrays that make up the cluster,” he said, “there’s no data movement. Yet all management is centralized. The beauty of this design is that with no network between the arrays to slow things down, performance truly scales with capacity and with consistent latency.”

The platform’s building blocks are ExaFlash Arrays, all-flash systems that Nimbus said are precision-engineered to eliminate irrelevant components and minimizing CPU utilization through “intelligent hardware-offload engines.” They also incorporate ultra-dense MLC flash drives for up to 50x the capacity, they are available in sizes ranging from 50 TB to 4.5 PB and include 32 Gb Fibre Channel, up to 100 Gigabit Ethernet, and up to EDR (100 Gb) InfiniBand connectivity.

“This is a direct assault on traditional hard drive arrays, and all existing all-flash arrays, for a broad range of applications, from virtualization and databases to cloud infrastructure, digital content repositories, and big data,” he said. “Worldwide storage demand continues to escalate, and with it, the need to increase data center efficiency, achieve scale with consistent performance, maintain simplicity, and reduce cost.”

Meanwhile, Netlist announced its HybriDIMM Storage Class Memory (SCM) product that the company said unifies memory and storage into one module (DIMM), “breaking the decades-long limitations of the Von Neumann computing architecture.”

Using an industry standard DDR4 LRDIMM interface, Netlist said HybriDIMM is the first SCM product to operate in current x86 servers without BIOS and hardware changes, and the first unified DRAM-NAND solution that scales memory to terabyte storage capacities and accelerates storage to nanosecond memory speeds.

"The promise of SCM is to bridge the price-performance gap between DRAM and NAND,” said Jim Handy, principal analyst, Objective Analysis, who said the SCM market could exceed $2 billion by 2020. “Others are attempting to cross this chasm with new proprietary materials. By combining the best attributes of widely available, proven technologies – DRAM's speed and durability with NAND's capacity, persistence and low cost – Netlist is able to deliver a cost-effective SCM product today."

HybriDIMM's architecture combines an on-DIMM co-processor with Netlist's PreSight technology ─ predictive software-defined data management – to unify memory and storage at near-DRAM speeds, according to Netlist, reducing data access latency by up to 1,000 times and provides up to 80 percent cost reduction.

"In today's data-saturated world, the challenge is in turning Big Data into actionable insight,” said C.K. Hong, chairman and CEO of Netlist. “By delivering the industry's first plug-and-play SCM solution, Netlist is enabling customers to process Big Data cost-effectively in memory and get to insights faster."

Netlist also announced agreements with Samsung and Integrated Device Technology. The company said volume production of the product is planned for the second quarter of 2017.

EnterpriseAI