Data Flow Acceleration and Processor Offload
The handling of information in the form of data stored and retrieved has received new attention with solutions to accelerate the data flow for storage and offload that function from the processor in servers and storage systems. Advances in other areas have enabled faster access to data: NVMe protocol to devices, solid state storage devices, NVMe protocol between servers and storage, faster internal busses for servers, and faster networking interfaces. These technology gains all improve the access to information and enable greater value to be gained from the investment in technology.
However, when an improvement is made in one area, it exposes the next area where there is a limitation. This ‘bottleneck’ becomes a focus for improvement and, depending on the economic and technical feasibility, becomes the basis for new developments with varying approaches. The latest technical advances have exposed the next bottleneck to be in the ability of the processor to handle the storage and retrieval functions for data, where previously it was with the interface protocols and device access.
This is the case now with several different approaches that address this bottleneck as well as providing other improvements deemed valuable. The variety of approaches available indicates there is more than one way to address this bottleneck. These are mostly recent applications of technology that are not in extensive use yet. As the value becomes more apparent and the awareness of the availability of the technology increases, data flow acceleration and processor offload solutions will increase in deployment.
Vendors covered: Fungible, Liqid, Mellanox, NGD, Nebulon, Pavilion Data Systems, VMware
Download now to read this free Technical Insight report.