Cloud vs. on-premises is about shaping risk

TL;DR: moving to the public cloud consolidates many small risks into few large ones. A recent GCP cloud outage (https://status.cloud.google.com/incident/cloud-networking/20004) highlights an issue with running more and more parts of the internet on the same infrastructure: as more computing resources consolidate in large cloud providers' data centres, the risk of a significant, global outage of … Continue reading Cloud vs. on-premises is about shaping risk

Varnish, session_linger and dropped connections

Recently I conducted a load test on my beloved low end Ubuntu 8.04 VPS. The setup consists of a mysql 5.1 DB, a Tomcat 6 webapp, openjdk 6 and Varnish 2.1.2. The load generator was issuing HTTP GET requests through a DSL connection which were supposed to create a decent load on Tomcat. To my … Continue reading Varnish, session_linger and dropped connections

A counterintuitive relation between CPU and memory usage

Continuing my tests on low-end server environments, I recently set up a stress test of a load balanced tomcat webapp. Part of that tests consist in reducing the available memory for the webapp (JVM -Xmx switch) and see how low it could go before negatively affecting performance. It's common knowledge that once available memory runs … Continue reading A counterintuitive relation between CPU and memory usage