Deploying in-network caches in support of distributed scientific data sharing

Abstract

The importance of intelligent data placement, management, and analysis has become apparent as scientific data volumes across the network continue to increase. To that end, we describe the use of in-network caching service deployments as a means to improve application performance and preserve available network bandwidth in a high energy physics data distribution environment. Details of the software and hardware deployments, performance considerations, and cache usage analysis will be described. We include thoughts on possible future deployment models involving caching node installations at the edge along with methods to scale our approach.