Optimisations on AWS Elastic Cache (Redis) Usage

Abstract

AWS elastic cache (Redis) is one the most popular mechanism for cloud developers to achieve both excellent caching as well as persistency of data. As part of our recent project of building personalised ad decision system, we needed to support stringent performance requirement of 125000 requests per second and 95% of requests must be served under 250ms. We used Elastic cache (Redis) for both caching and persistency and found number of optimisation techniques which helped us to achieve the performance requirement. These techniques we used ensured that cost for Elastic cache (and the whole project) is much lower than initial estimate. The techniques involved choosing appropriate CPU instances for Elastic cache, using right cache structure and whole host of changes on Redis client side to make this end-to-end system high performance, and cost effective.