Upon further monitoring, everything seems to have stabilized for quite sometime now.

We've placed measures in place which should prevent the increased latency.

We're currently doing an emergency maintenance which will result in slower than expected I/O. This is in order to resolve the issue seen over the past few days.