Virsto's software-only solution to tackle storage problems in virtual environments must work. Had it not performed, investing in a XenDesktop version would not make sense, because virtual desktop infrastructures (VDIs) pose greater performance challenges even than virtual servers.
When I first reviewed Virsto last autumn I saw a promising "up-start" proving its solution on Microsoft Hyper-V with a VMware ESX implementation due a couple of months later (since released). Last week, barely six months on, I talked to a company with around 100 customers and now a XenDesktop version.
The Virsto deploy wizard supports XenDesktop 5.5 on VMware ESX/ESXi 4.1 and ESXi 5.0. (Virsto already supports vSphere VDI and Hyper-V VDI.)
So what are the VDI performance issues? First, VDIs tend to generate more write contention than virtual servers through having many more virtual machines (VMs) per host. Then, there is a much greater fluctuation between normal and peak traffic because of simultaneous boot, login, application loading and logout situations at the start and end of the day ("boot storms"). Add to this a greater difficulty in provisioning correct capacity per desktop and, for Windows virtual desktops, the need for a dedicated local disk.
Virsto installs on VMware as a virtual appliance that looks like a virtual machine disk (VMDK); this produces fixed performance and thin-provisioned VMs, with each VM having a Virsto virtual disk (vDisk). This process bypasses the native VMFS. Virsto's VP of marketing, Gregg Holzrichter, politely described VMFS as "not optimal" and said that the space-efficient Virsto vClones use up to 90% less storage capacity with best performance. "Virsto achieves the highest performance bar none", Holzrichter claimed. (I previous described some of the other technology aimed at fast throughput and easy deployment.)
The proof is in the pudding of course. As the number and size of users grows, figures will prove that either way versus the competition and the extent to which that is true for each hypervisor Virsto now supports. I also expect implementation of its first seven-figure order (a so far unnamed client) to provide some telling performance figures.
Related performance claims for Virsto on XenDesktop include an expected doubling of the number of desktops per host, an increase to 90% storage utilisation, Virsto snapshots and clones using native workflows to make provisioning and backup easily scale, and all with transparent VDI management.
So why the VDI market push, as the VDI market is smaller than that for servers? Holzrichter agreed that the VMware server storage market remained its biggest opportunity, but pointed to some user research that showed 67% of organisations asked planning to go for VDI usage in the coming year. In other words, he saw 2012 as the real take-off year for VDI. Virsto is now well placed to exploit this.
A lot of this is about saving money (CAPEX and OPEX) - for instance through minimising admin staff and investment in new hardware, which is what Virsto is about. So far, Virsto's expansion has been overwhelmingly in the US, as development took priority over widening its markets. Now Virsto is beginning a European push - and potential cost-savings will surely be a high priority here.Â