Consolidation addresses the business needs described above by bringing together a number of physical servers into fewer physical host servers. In this way resources can be utilised more efficiently, Disaster Recovery & High Availability strategies can be more simply managed and hardware failures can be more effectively mitigated. Specifically, the business needs are addressed as follows:
Current Virtualisation products, such as VMware, provide a way of combining a number of physical servers into a single physical host. Within the host, the servers run as ‘Virtual Machines’, unaware of their new status, and appearing to the external IT infrastructure just as before in terms of networking and functionality. In this way, the processor and memory utilisation can be managed such that it achieves, ideally, an average efficiency of around 80%. This allows IT managers to make more efficient use of their expensive computing resources.
The benefits of consolidating several servers into a single host can be further enhanced by grouping two or more hosts into a resource cluster. When combined with a SAN, NAS or iSCSI storage solution; VMware can dynamically balance the placement of virtual machines across host servers. This balancing process can be completely autonomous and can be performed without losing the services provided by the individual virtual machine.
As discussed above, by grouping two or more hosts into a cluster linked to a SAN, NAS or iSCSI storage solution; VMware can implement High Availability (HA) across the cluster. In this way, should a host server fail; virtual servers that were previously running on the failed server can be automatically restarted on another host server.
When coupled with a traditional clustering solution, such as Microsoft Cluster Services; levels of HA can be achieved that provide ‘five nines’, or 99.999% levels of availability.
The usual response from IT managers, when invited to consolidate physical servers into a single box, tends to involve the phrase “all your eggs in one basket”. This is a reasonable response and is discussed further below.
Whilst this process may at first glance appear counter-intuitive; the eggs-in-one-basket scenario can be addressed successfully in two ways:
By requiring fewer physical servers (VMware will support between four and eight virtual servers per physical CPU); it is possible to invest in higher specification servers that are, in turn, able to provide higher levels of performance and fault tolerance. By deploying physical servers with redundancy in their power supply, memory and networking components; it is possible to mitigate the likelihood of a single server failure at a reasonable cost.
VMware High Availability (HA)
As discussed above, VMware is able automatically to restart a failed server on another host server when it detects a failure of a host. This process can take as little as 30 seconds; however implementing a complimentary high availability solution (such as Microsoft Clustering Services) can reduce the downtime resulting from a physical failure to zero.
To find out how Consolidation can help you – please contact us on 0161 848 4315