Edge Computing Analytics on Computational Storage with Resiliance and Fault Tolerance using VMware VSAN, Bitfusion and Tanzu
NGD Systems, Inc.
Category
Hyper-convergence Innovation of the Year
Entry Description
Emerging edge use cases need to process vast amounts of data and yield AI-driven insights faster than ever.
These systems are increasingly in the critical path and need to handle host failures without a hiccup for mission-critical environments.

This Innovation explores a new solution that combines leading computational storage technology from NGD Systems, a VMware Tanzu Greenplum database and VMware vSphere 7 for a high-performance, zero-downtime edge analytics solution. It enables computational storage with GPUs by using vSphere Bitfusion technology, allowing efficient machine learning inference and even training new models at the edge. VMware vSAN and vSphere Fault Tolerance work together to offer zero-downtime edge analytics for a small form factor deployment.

The net result of this innovation is a platform that can replace a rack level (42U) GreenplumDB solution with a Edge-Friendly, Fault Tolerant, 6U platform.
Overall results of the analytics are faster than the traditional architecture and provides a better overall TCO and use of the system components.

This platform requires only 3 single socket servers, 3 GPUs, and full array of Computational Storage Drives to accomplish like task that would normally require dual socket servers with 16 GPUs and empty storage slots (as it requires more servers for CPU/GPU) not data.
Supporting Documents