Over the last few weeks we have been working and planning to restructure our network for the first quarter of this year.
Back in early 2016 we made the decision to migrate sites from physical servers to virtualised servers in order to try and provide additional resiliency and redundancy from hardware failures. Since we were creating servers inside servers, if we detected a potential hardware issue we could simply move these containers from one physical box to another. While we are happy with this implementation, we feel that it now adds an unnecessary step of complexity to our infrastructure and provides very little gain. Hardware failures, while they do indeed happen, are relatively rare and the virtualisation is perhaps best suited for other tasks such as the isolation of separate environments. In addition to the removal of the virtualised servers, we felt it was time to also perform a hardware refresh. A hardware refresh is essentially an upgrade of hardware – for example, we may utilise faster and more power efficient CPU’s, or increase the amount of RAM in a server. As technology changes every few years sometimes it is beneficial to take advantage of the latest and greatest technology. With these changes we will be ensuring that all of our servers now maintain; a minimum of 128GB RAM, a minimum of 12-16 CPU cores and pure SSD drives with at least 1-2TB of storage capacity on each server.
Because of the increased specifications, we are also deciding to merge semi-dedicated and shared/reseller hosting packages on the same servers – but will continue to provide higher CloudLinux and resource limits to these semi-dedicated packages. This will allow customers to upgrade (or downgrade) from or to a semi-dedicated if the need arises, without any interruption to their service at all. Previously we would have needed to manually move an account from one server to another which would result in some brief interruption while DNS propagated.
Hardware aside, we have also decided to use different network providers in order to hopefully improve performance for all of our customers. On our UK server side, we have noticed that some of the routing paths taken by our upstream provider are quite poor, and can result in increased latency and decreased performance to an unacceptable level during peak hours. We have been looking for a new provider in the UK and have located a great match with a superior network to our present one. Over the next few months we will be migrating our UK customers over to this new network and hope that our customers will see an improvement. On the US side we have done much the same and are already part-way through the process of migrating customers.
We understand these migrations can be disruptive for our customers, and for that we sincerely apologise. We are committed to providing the best service that we can, and in order to do that we sometimes have to make significant changes such as this. We feel strongly that these changes are important to provide you with an improved service that you will be happier with.
If you have any questions, comments or concerns regarding this blog post, then please feel free to post in the comments below. Alternatively you may open a support ticket from our client area to speak to a member of staff.