Some websites and applications need a higher level of uptime than can be realistically achieved in a traditional VPS or Dedicated environment. For these high availability websites we have developed our VMware/Hyper-V Cloud Servers. A cloud can only be as good as the hardware and software it is built upon. With this in mind, we opted to use quality branded Dell PowerEdge R620s in our latest VMware/Hyper-V clusters. Each server is powered with dual quad core CPUs and 96 GB RAM to ensure each hardware host can handle the needs of the VMs residing on it. In our clusters we keep two hardware hosts available in case of a physical failure on another host. In the event of a failure the VMs residing on the failed host will be automatically started on another physical hardware host. This ensures that in a hardware failure the downtime experienced is literally the time it takes to reboot a server.
The key part of a cluster is undoubtedly the remote storage devices in use. It is relatively easy to build a cluster with hardware failover, but to do so while giving comparable performance to a dedicated platform takes experience in both the software setup and also in depth research in choosing the hardware. Environments from lesser hosting companies are plagued by slowness from storage arrays which have been massively overloaded. Our storage cluster is built upon EMC Isilon SANs to give maximum performance and reliability. To give an extra speed boost, we have added SSD to each SAN for use with Meta Data acceleration. We also have three different caching patterns which can be utilised in order to optimise disk layout by workload. By doing this we can ensure that virtual servers with a light to medium disk I/O requirement are served from a file system optimised for this purpose. At the same time, for those with high I/O requirements, such as database servers, we can move these virtual servers to a file system optimised for this purpose.
Building a highly redundant cluster would be somewhat pointless unless the facility it is placed in is also highly redundant. We house our VMware Clusters in a Tier 4 facility, rated N+N for network and power, to ensure the highest possible levels of reliability. The facility is staffed 24x7 by experienced NOC engineers to ensure our environments are proactively monitored and managed. Because we are so confident in our platform, we offer a 100% uptime SLA in the event of either the hardware or software not functioning as expected.