04th May 2020 - 11:00am

With the Covid-19 pandemic the stress on resources has increased significantly, the network throughput and disk demands have literally gone up by 5 times in some cases, this has highlighted a number of issues one of which is balancing, as you will appreciate it is impossible to know which customers will make which demands on resources when and in a multi tenant shared resource environment that can cause performance impacts.

This has been very well managed over the last 10 years however there is no way to plan for entire nations suddenly working from home which has made this a bit harder to handle.

A few nodes have some bad combinations of qemu, libvirt and kernel versions causing sub optimal performance for NVMe especially, this can only be resolved with a reboot however with the lack of remote hands within the data centers for obvious reasons right now the preference is not to send people to site unless it is a true emergeancy.

Initially it was thought that the lock down would be over by mid to late April obviously that has not happened and now it is looking like it could even stretch as far as July.

As such the best way to handle this without potentially causing long term mass disruption and forcing people in to bad situations is to rebalance some services.

If you are on NewKVM2, KVM5, KVM6, KVM9, VZSSD25 or VZSSD26 (you can see this in the VPS management page) the over the next 24 - 48 hours there will be additional disruption while services are moved around to better balance the demands.


Update: 7th May 2020 - 09:30am

Further balancing will be taking place for NewKVM2, KVM5, KVM9 and KVM6 over the next 24 hours

Monday, May 4, 2020

<< Geri