Mendel delivered the keynote session today – the topics were vService and vSecurity. Mendel did make a node to Storage VMotion and Continuous HA. I know, Continious HA was shown last year at VMworld – but I was just hoping they would do it again. For the benefit of those people who didn’t get to go VMworld, and don’t have the energy to visit VMworld.com. Mendel seems to come alive with the technology. He has that little cheeky grin of excited schoolboy about him with his product demos rather than fear it might all go horribly wrong. However, you can see how kinda tired he is with the “establishing a CTO style business case for product A”. I guess you could say that’s the sign of real technology officer/professor in a University for you. So what is vService and vSafe. vService used the OVF (Open VM format) to download groups of VMs which reflect a distributed service. It’s a next, next, next approach to say rolling out a new CRM (demo’d) or any large suite scale application service – say like a full remote office function like Citrix or email solution like Exchange/OWA. My only scientism around this was that it was suggested that this download process (which is all it really is) could be finessed by including the “metadata” – or in less fancy way the configuration inputs (your configuration specific to your environment) could be collected along the way. This is really a huge undertaking – if you take beyond changing IP address and such like – to including the specific info required to complete the configuration and installation of the whole Citrix Presentation/Metaframe/ZenServer/AppServer (what shall we call it this year, guys) system.
In contrast vSafe seemed like a lot more realistic undertaking. Basically, opening up the hypervisor to allow 3rd party security suppliers (McAffe took over at this point) which could both anti-virus and intrusion detection outside the VM but inside the hypervisor. Of course, most companies already have deployed physical appliances at the perimeter of their networks – spam, antivirus, vanilla intrusion detection systems. So I think it might be sometime before the switch over. It remains to be seen what the performance implications of such a model might be at such an early stage. I imagine some people might have concerns about the implications of the VMkernel becoming much greater than simply being a super-thin, super-reliable hypervisor.