During the scheduled release of an update we inadvertently left the application pool with insufficient capacity for the level of usage we were experiencing at the time. This resulted in slow behaviour and partial query timeouts for a period of around 10 minutes whilst the code was deployed and the system capacity stabilised.