Over the last several years XenServer competitors have presented several arguments against using XenServer, and often those arguments center around a feature they had which we didn’t.
A perfect case in point was the issue of virtualization ROI and memory sharing. VMware was often ready to present that vSphere had better ROI due to their abilities to balloon, share and swap memory. Despite issues in methodology of such reports, this was a fair argument when XenServer didn’t have the ability to manage VM guest memory. However as soon as we introduced our memory ballooning solution, Dynamic Memory Control, the focus shifted away from memory as the key factor in ROI, and storage live migration became the new focal point.
There are of course a number of fairly common scenarios under which migrating a disk from one storage solution to another make sense, and prior to XenServer 6.1 it was pretty easy to satisfy those requirements; so long as you didn’t mind turning the VM off first.
The interesting thing is that when you are talking about highly available applications, assuming these applications are deployed per best practices, they will keep running with one node offline; yet administrators were reluctant to incur the downtime. The number of arguments put forward for why live storage migration was a true requirement did impress me, but when you think about it this was an effective marketing effort creating a need to match the capabilities of a feature.
With XenServer 6.1, we too have addressed the same need for live storage migration, but also recognized a key component of virtualization today; successful datacenters aren’t bound by arbitrary cluster sizes any longer. Successful datacenters choose a clustering size which makes sense for the types of workloads being hosted, and if you really need live migration, you are by definition going to need live migration across clusters. With Storage XenMotion in XenServer 6.1, that was one of the first design requirements. This means that whether you have resource pools with a single server, or resource pools with only local storage, or hundreds of resource pools, Storage XenMotion was designed with you in mind.
Of course, live migration of VMs still didn’t address the more traditional enterprise use cases like replacing aging storage arrays or addressing a LUN which was running out of space. In these models the VM itself stays stationary, but the underlying disk is the piece which moves. Thankfully by addressing a core requirement of large scale datacenters and cloud providers first, we have also created a solution which was perfectly suited to deal with the operational requirements of storage management in enterprise datacenters.
If you want to see Storage XenMotion in action, please take a look at this short video series I recorded which covers three common usage scenarios. Better still, if you want to experience Storage XenMotion for yourself, download a copy of XenServer and then start a free XenServer Advanced Edition trial. If you are looking for a little light reading on the subject, please take a look at this overview paper. With almost a million downloads of XenServer under our belt, we see that when IT wants to balance functionality with cost; XenServer is their clear choice and with Storage XenMotion that choice just got a bit easier.