Follow Datanami:
October 26, 2017

IBM Stacks Memory Cells to Boost Flash Capacity

All-flash platforms are steadily being used to drive down the cost and infrastructure required to store soaring data volumes as more enterprises shift to hybrid cloud infrastructure. Recent technology advances involving the stacking of flash memory cells have greatly increased storage density while continuing to meet data scientists’ performance requirements.

With an eye toward upgrading datacenters to support hybrid configurations that could be use to extend data access across public and private clouds, IBM this week rolled out advances in flash array storage that among other things would provide a three-fold increase in the provisioning of storage capacity in the same physical space.

The company (NYSE: IBM) also unveiled virtualization software designed to ease the migration of data to and from public cloud services along with storage options that work with emerging tools like application containers and data applications such as machine learning. The company said the datacenter storage upgrades also would support both Docker and Kubernetes application containers.

Along with lowering storage costs, IBM said Thursday (Oct. 26) its flash arrays and storage software are intended to make data more accessible whether it is stored on-premise or in the cloud. The company promoted is FlashSystem array as reducing “data capacity costs” by nearly 60 percent.

The new flash array also in being incorporated into an upgrade of VersaStack converged infrastructure IBM jointly offers with Cisco Systems (NASDAQ: CSCO).

Andy Walls, CTO of IBM’s FlashSystem, said the flash storage market hit a wall as scaling of memory cells went to 16-nanometer features sizes. “The floating gate became so small that I knew every electron by name,” Walls quipped. “We needed to do something different, and what we did differently was instead of making the cells smaller we said, ‘Let’s add layers to the flash technologies’.”

The result is three-dimensional stacks of memory cells that are three times denser than previous cells but housed in the same datacenter footprint. IBM claims the 3-D approach yields 180 terabytes of maximum storage capacity with “consistent performance,” meaning “low latency all the time,” Wall said.

As flash cells expand, so too did bits-per-cell capacity, from 2 to 3 bits. That density advance, dubbed 3-D TLC, resulted in another 50 percent boost in storage capacity. Wall said the memory cell stacking technology is being incorporated across IBM’s FlashSystem product line.

Advances in solid-state arrays by IBM and other flash storage leaders such as Dell EMC, Hewlett Packard Enterprise NYSE: HPE and upstarts like Pure Storage (NYSE: PSTG) are expected to double storage density and reduce storage costs over the next year, according to market watcher Gartner Inc. (NYSE: IT). Demand is being driven by enterprise customers looking for ways to store and access huge volumes of unstructured data from sources like machine-generated log files, social media feeds and Internet of Things sensors.

Hence, data scientists are seeking are increasingly turning to scale-out flash storage as a way to facilitate new analytics stacks.

IBM said the new features to its all-flash systems and storage software would be available by the end of the year.

Recent items:

Supercharging Apache Spark with Flash and NoSQL

New Flash From IBM Targets Unstructured Data Analytics

 

Datanami