Once, apps and OSs were running on dedicated physical servers. The problem of such setup is that utilisation of mighty and expensive processors is 5%. The rest is unused. The utilization at best goes up to 20%.
Also, if you want to upgrade your hardware, the system has to be shut down, resulting in downtime.
The solution to these two problems is to separate hardware from the OS and apps, and introduce another layer, the so called virtualisation.
The pioneer of this is VMware. The OSs running on the virtual platform don’t know the physical machine they’re actually on. The OS simply doesn’t know.
The virtualisation layer can be done in two styles.
One, is the host-based virtualisation system. It requires an OS as it basis. This is not ideal.
Two, is bare metal hypervisor system. As the name implies, the virtualisatin goes directly on top of HW, no OS needed in between. This is in vogue now.
VMware offers its complete virtualisation package under the market name of vSphere, currently at version 5.0.
The first block of vSphere is the HW virtualisation layer, ESXi. The above described virtualisation at its essence is this ESXi, in essence VMs, the virtual machines. To control your group of virtual machines you need vCenter Server. vSphere Client is the client side access to vSphere. There are some other ingredients to the system. Important, yet not for this intro.
From the user’s perspective the virtual machine is indistinguishable from a classical machine. From the perspective of the hypervisor, the virtual machine is simply a stack of files.
This brings forth several advantages. Moving VMs from one physical server to another server is very easy, it only requires copying of appropriate files. Any changes to underlying HW does not affect the VM at all. Yet, the improvements to the underlying system trinkle up to the user nevertheless.
On the side of utilisation vSphere enables you to setup resource pools. Within the resource pools VMs compete for resources. Whenever a user needs more resources it can grab them from the dedicated resource pool. If all users within a pool of a hardware machine, the machines can be seamlessly moved to other unused machines in the datacenter. One step more, when a physical server breaks, VMware moves it automatically to another working machine within the datacentre. Another step, when people go to sleep at night VMware realises that too many physical servers are running, so it moves all VMs to a few or even a single machine, and it shuts down all idle hw machines, until people start coming back in the morning.
Another advantage is consolidated backup. vSphere takes care of a single backup routine for the whole datacentre instead of multiple backups on each of the separate machines.
Now, what are the business effects?
At NIL we run 400 VMs on 4 UCSs. And they are running at about 50% capacity right now. Cost savings in HW alone are in the range of 80%. It is true that a substantial part of the savings goes back to VMware for its magic. It is also true that this stuff ain’t for your Sunday tourist IT wannabe manager. But, again, how much more manageable does this become on the side of your sheer organization. The verdict is clear, VM stays here, and with it the Cloud.
At NIL we call our private cloud the HyperCenter, our way of designing everything, from HW, to virtualisation, to clients, to storage, to networking, to policies, security, …, the whole IT shebang. It is designed for highest demanding service provider and enterprise level needs. Some of the biggest corporations in Slovenia and worldwide have started using it. Should you feel like you wanted to join them, call us, send us a mail at firstname.lastname@example.org. Or come to our 1-day introductory seminars. The first one, to be held in Slovenian is on August 30th @NIL HQ in Ljubljana.