HubSpot engineering is a Java shop invested heavily in microservices and continuous deployment. Java is not only used to run our thousands of deployables, but also our queues (Kafka) and Big Data solution (HBase). Keeping all these JVMs performant while providing a good user experience has forced us to dig deep into Garbage Collection (GC), particularly the Garbage First Garbage Collector (G1GC).

An interesting task I undertook recently was hardening our memcached infrastructure. We're an SOA shop with hundreds of deployables, most of which will interact in some way with our memcached clusters. I thought I'd take the opportunity to break down the testing process into a basic tutorial that could be repurposed for other external resources.

A long time pain point for HubSpot Engineering has been intermittent core dumps with Hadoop jobs.