Xen Orchestra 4.2

All ahead full! 2 releases per month is now our current pace, and it doesn't mean less features each release. As you will see, this one delivers some interesting new possibilities, combined with a huge performance boost.

Schedule rolling snapshots

The first step toward backups: you can now plan snapshots in a very natural way. Let's see this step by step.

Imagine you want, for some reasons, to be able to revert a critical VM to it's previous state during the last hour, up to the last 4 hours. And only from Monday to Friday. Complicated? Not at all.

You can do it in few clicks now, as you can see in this video:

You'll have:

  • a snapshot every hour
  • with a maximum of 4 snapshots (the old one is removed automatically)

This feature is very flexible, you can:

  • schedule a nightly snapshot for a group of VM, let's say at 4:30 in the morning. With a max snapshot at 7, you'll have a week of revert possibility
  • schedule a snapshot every week, but precisely Sunday at 11 PM, with 4 snapshots max. This will let you a month max of revert
  • and so forth!

Huge performance boost

You can read more in our previous post about how we managed to boost Xen Orchestra up to 250 times faster in big infrastructures.

On our small lab we already got a 4 times faster XO. See for yourself, lower is better, 4.1 vs 4.2:

And the other good news: it scales well! Bigger your infrastructure, better the boost!

In this kind of infrastructure, XO is now blazing fast!

Compatibility with 5.x versions

The new event.from system we are using is only working since XenServer 6.0. But some people are yet using older versions! Thus, we created a fallback on event.next: it works now.

New VM creation process

We completely re-wrote the VM creation process: it's more maintainable and less prone to bugs.

Bug fixes

You can check the change log for all the details.

Upgrade!

Just use the embedded upgrade process in the XO interface. Nothing else to do :) Refresh your browser after the update, to be sure you got the last version.

Next?

We made our first step with scheduling snapshots. The next step is clear: provide a scheduled full export of VMs, probably in a NFS share. Disaster recovery is probably the first thing you got in mind with that.