Dennis Waldron, March 2015 –
At Plumbee we take a data driven approach to analysing our business and hold strongly to the belief that you can never have enough data. As a result a significant proportion of our workforce are dedicated to the analysis of events from sources as diverse as Customer Relationship Management (CRM), advertising, database writes and Remote Procedure Calls (RPCs). On any given day we collect approximately 450 million events adding 166GB of compressed data to our existing warehouse of 87TB, all of which is stored in Amazon’s cloud based storage solution S3 (Simple Storage Service).
In a previous blog post we discussed our approach to logging analytics events represented in JSONinto Amazon’s Simple Queue Service (SQS), a highly scalable, reliable and fast cloud based message queue. In this post we cover the consumer side of that story focusing on how we process and store those events in S3 using Apache Flume NG.