Attala's Hardware-Based Storage Makes a Splash
Newcomer Attala Systems Inc. made a big splash at the recent Flash Memory Summit, collecting the Most Innovative Memory Technology Best of Show award.
The 2017 Flash Memory Summit was held in Santa Clara, Calif. While there, Attala staged a demonstration with Intel, using the chip giant’s FPGAs.
Attala’s technology employs scale-out fabric that leverages Ethernet networks to interconnect servers and data nodes in the data center. It can support up to 10 million IOPS per scale-out nodes and keeps latency as low as 16 micro seconds.
This automated solution enables the fabric to attach storage resources across the network, directly to where the application resides. That could be in a bare metal server, a container, or a virtual machine, the company explains.
But while the themes of automation, bare metal, containers, and VMs are frequently associated with software-centric networks, Attala emphasizes that its hardware design is the important thing to consider in this solution.
“Hardware-based data-paths and queues – that are flawlessly predictable – are combined with hardware-based QoS controls and granular monitoring – that have zero performance perturbation – to create an end-to-end solution that enables workloads [to] realize predictable low latency, IOPS and throughput while maximizing the efficient use of a shared infrastructure.”
At the event in Santa Clara, Attala demonstrated its High Performance Composable Storage Infrastructure reading data with round trip latency of 16 micro seconds. That, the company said, is an industry first for networked storage.
“The way we achieve such low latency is because all the data paths and scheduling between the application and SSD are 100 percent in hardware,” the company explained in an Aug. 6 blog. “There are no software interrupts, polling, scheduling, memory copies nor other miscellaneous software layers getting in the way. The same infrastructure is what enables predictable and consistent performance regardless of whether you use Optane SSDs or more affordable 3D-Nand SSDs.”
Edited by Maurice Nagle