HW Accelerated Key-Value Store as a Service

Mon Sep 16 | 4:35pm
Location:
Cypress
Abstract

AI and ML storage use cases are increasing storage demands. Key-Value is the fastest-growing API for storage and underlies many database and Web-scale applications. Storage computing needs to move from general-purpose CPUs to more efficient, lower power, and faster accelerated compute platforms. Storage is lagging in this area but however, is consuming significant CPU resources. We will discuss the challenges of this evolution and how the technical issues may be solved.

Learning Objectives

Upon completion, participants will understand the fundamental challenges a storage solution must address and gain insights into the current state of legacy solutions, with a focus on Key-Value storage systems.
We will analyze the power, performance, and cost issues in legacy storage solutions, identify their root causes, and explain how a hardware-accelerated solution can effectively address these problems.
Upon completion, participants will understand how to implement an accelerated solution within existing cloud infrastructure and how this integration will transform the storage infrastructure.


---

Andy Tomlin
QiStor
Related Sessions