SC19: Collaboration of Eideticom and Los Alamos National Laboratory

Eideticom Communications, Inc. and Los Alamos National Laboratory announced the results of a collaboration to develop NVMe-based compressed parallel file system.

This high-performance Lustre/ZFS-based parallel file system leverages Eideticom’s NoLoad, a NVMe-based Computational Storage Processor (CSP) that offers performance and efficiency benefits for HPC. The collaborative effort was sponsored under LANL’s Efficient Mission Centric Computing Consortium (EMC3).

NoLoad CSP is the storage industry’s first NVMe compliant scale-out accelerator in a U.2 form factor. The CSP accelerates ZFS by offloading key storage services and utilizes energy-efficient FPGAs from Xilinx and BittWare, a Molex company.

The News:

• NVMe-based computational storage compressed parallel filesystem built using NoLoad CSP and deployed in LANL’s Lustre/ZFS-based HPC parallel filesystems.
• NoLoad CSP’s high performance compression engines provide scalable offload of storage centric services and enable capacity increases with no impact on performance.
• NoLoad’s NVMe-compliant interface simplifies deployment of computational offload by making it straightforward to consume in servers of all types and across all major OSs.

“We are excited to see standards-based computational storage technology being applied to a growing, acute problem in data movement, namely storage server memory bandwidth,” said Gary Grider, deputy division leader, Los Alamos National Laboratory’s HPC division. “Offloading these memory bandwidth intensive operations to a standards-based NVMe Computational Storage device is a perfect initial use case. We have enjoyed a great working relationship with Eideticom, who is leading the charge in NVMe computational storage and look forward to our continued partnership under the EMC3 collaboration program.”

“We are entering an era with fast storage systems that pair high-speed NICs with fast NVMe storage,” said Brad Settlemyer, senior scientist, Los Alamos. “Memory bottlenecks make it impossible to perform critical storage system tasks such as compression, erasure, check-summing and dedupe at the rates supported by the storage and networking devices. The Eideticom NoLoad devices have demonstrated that we can offload storage functions onto accelerators enabling line-rate compression, improving CPU utilization, and reducing memory bandwidth pressure.”

“We are excited about the results of the collaboration with Los Alamos,” said Roger Bertschmann, CEO Eideticom. “The combination of LANL’s storage system expertise with Eideticom’s NoLoad CSP resulted in the achievement of a world best ZFS benchmark with inline compression. The combined solution delivers all the performance of NVMe storage while achieving the cost-saving benefits of best-in-class compression.”

Eideticom was demonstrating new products and end solutions at SC19 in Denver, CO, on November 18-21 with partners Xilinx and BittWare.