|
|
|
|
VMware 2:
Taking back time...
and adding something free into the bargain
Whilst Exchange may be a memory hungry beast and backing up terabytes of data does tend to test the performance of a SAS array, for the majority of the time, as most sysadmins will testify, servers are idling along, happy to be using no more than a small percentage of their performance capacity.
For some time now, 8Networks has been well aware that rolling out a fleet of Dell PowerEdge 2900s just to cope with these ‘burstable’ demands on a server’s resources is a teeny bit overkill. So in-house we’ve been running various virtualisation products on top of our ‘big metal’ to maximise average utilisation on our servers whilst still providing room to breathe, so to speak.
As time marched ever on, and evidently virtualisation came to the forefront of every IT bod’s mind, we decided to look at passing the benefits of virtualisation on to our clients. The technology was proven; we’d been the guinea pigs and had successfully ‘burnt-in’ several deployments of differing server technologies on top of hypervisors, ironing out the creases as we went.
However, whilst we were confident the technology worked and proved invaluable to us, we weren’t so confident that the hefty price tags involved with enterprise-level hypervisors would be met with big smiles and signed cheques by our clients with a 15 workstation network.
In stepped ESX’s little cousin (I say cousin purposefully, as will be revealed)—VMware Server. Provided free by VMware for businesses to dip their toes into the world of virtualisation, this would allow our clients to make use of the benefits of virtualisation without the dreaded four figure price tag.
Okay, they won’t get some of the cooler features provided by the likes of ESX, for example VMotion et al, nor will they receive support from VMware—but we’re on hand for that. On paper, it was perfect—it allowed clients to run multiple hosts on top of a single server, vastly reducing costs involved with purchasing hardware.
Digging a little deeper, which we make it our job to do, there were of course ‘limitations’—only one processor could be assigned per host, and each host was limited to 3.6GB of allocated RAM. But for small clients, these limitations fitted within the boundaries of their needs quite comfortably.