Storing at the Edge - An Architectural Approach

Library Content Type:
Publish Date: 
Wednesday, September 26, 2018
Event Name: 
Focus Areas:

Edge computing has become increasingly critical as more endpoints. Edge computing is intended to optimize networked systems by processing data near the device generating the data. In reality, it is much more complex and requires intelligent decisions about storage technology at each tier of storage and transmission.

In this paper we take a close look at the workloads that run typically on edge and their IO pattern. Given the type of IO pattern and data volume that edge creates, we are looking at possible storage solutions that addresses the key parameters. Finally, we define a set of properties required for typical edge storage solutions that suites the emerging workloads targeted on edge.

Learning Objectives:
1. Edge computing and its needs
2. Typical edge workloads - ML, online learning. deep learning
3. Storage IO pattern - analysis
4. IO architecture

Watch video: