VMX Bandits

So this is pretty old news now (~1 mth?), but it may as well be said - there's an update to the SVMotion GUI plugin provided by the vip-svmotion.wiki.sourceforge.net/ project guys - I can't see any real difference with it to be honest, but presumably it's better in some abstract way. Or something.

Also, a newly installed esx3.5 server I connected to today asked me.. well actually held a gun to my head and forced me to download a new VI client. So far I can't really see any difference with this either, though perhaps it now lets ESX see LUNs that are formatted with bad ju-ju without taking down the whole ESX host. In any event, there is one aspect of it that's slightly worse. Nothing major, except it crashes and burns when I try and connect to VI with it. So not critical or anything.

"By default with VMware Server 2.0 Beta, debugging mode is enabled for all hosted virtual machines. This has been the case with every VMware beta product that I can remember. The problem with debugging mode is that it significantly slows down the performance of all on-board virtual machines so much as to make them unusable."

Head on over to Realtime Community for the answers.

"VMware Storage VMotion is a state-of-the-art solution that enables you to perform live migration of virtual machine disk files across heterogeneous storage arrays with complete transaction integrity and no interruption in service for critical applications."

Sounds a lot like VMotion itself, which lets you migrate live VMs around between hosts in your pool or recover transparently from host failures, except it's not quite what you might think at first. DRS balances your guests around to better utilise resources, and HA lets you recover quickly on a host failure. There's no equivalent of either for storage, and Storage VMotion doesn't do anytihng like failure recovery at the storage level, which kind of sucks. You are expected to manage your own storage redundancy, and this ends up costing usually a lot more than the rest of the system if you do it to the same level of redundancy and performance.

I had a rather large post ready which explained about how fiberchannel is too expensive, iSCSI corrupts data and is slow, and NFS is hard to configure for redundancy (and lets not even get into how annoying it is that ZFS as the backend FS for NFS makes VMware guest performance just plain suck to the point of unresponsiveness) but it's all rather pointless really. The problem is that storage is the base resource that everything else hangs off, so if the storage array goes away, everything falls over. This isn't an easy problem to solve either.

I guess VMWare could write to two or more storage arrays (perhaps you could add them to the pool like you do ESX hosts) and simply revert to using a subset of them in the event of a failure. Sort of like an application layer n-way mirror. There's some specifics to work out, but I can't see how it'd make it much slower than the slowest store in the pool (hell, for reads it'd have to increase performance, you'd think!) It'd have to be cheaper to buy a couple of NFS boxes attached to some RAID storage than buy a SAN.

So storage VMotion just lets you manually move storage around to perform maintenance on the backend storage etc (which is definitely a good thing, don't get me wrong) but doesn't really help when it comes to a storage version of the standard vmotion/HA.

Oh, and the lack of a GUI on Storage VMotion is IMHO inexcusable. Seriously VMware, what the hell.

"Tom Valovic over at Virtualization Review magazine pens an interesting (albeit short) comment about VMware's current positioning and how it resembles that of previous notorious companies like WordPerfect Banyan, and DEC."

Check out this article over at Realtime Windows Server.

More security issues to report, a vulnerability was reported in VMware in the VIX Application Programming Interface (API). This meant a local user could obtain elevated privileges on the target system. VMware ESX also has the same fate with it's vulnerability.

An issue was reported with VMware on Linux-based hosted systems and VMware Tools, where a local user could obtain elevated privileges on the Windows-based target guest operating systems.

Time to get patching ... but from what I hear, ESX isn't the easiest thing to patch -- VMware should really address this.