Performance seems to have stabilised, we'll continue to monitor over the next few hours.
This outage has shown that we don't have good enough monitoring in place - we'll work on putting something in place to catch these going forward. Apologies for the downtime.