You wet-behind the ears buzzword loving newbie admins make me laugh. To those that say virtualization is good for the purpose of redundancy if a server fails, I say you have no idea what you're talking about. Virtualization is a double-edged sword, and there's a reason IMPORTANT servers are very rarely virtualized.

First, a server very rarely goes down due to hardware. It requires a serious issue for that to happen, such as a failed motherboard or backplane. Those types of failures are incredibly rare. The majority of failures are just hard drives or fans. Hot swap affairs that requires no downtime. Even memory can be hotswapped on many servers - though memory chip failures fall into both categories; they are rare, and they won't take the server down. Memory is rarely installed in a single chip configuration, and in most cases servers have at least two banks filled, and sometimes even mirrored for redundancy. Sorry, college graduate-wannabe admins... Servers very rarely go down due to hardware.

In the rare event that your motherboard does fail, you can do the same thing with the physical hard drives -- just swap them into another box and be back up in a jiffy. Only if there is a huge difference in the hardware would it take more than a few minutes for the OS to reconfigure itself.

Think about this for a moment: if your system fails due to hardware, and it's down -- how will you copy the virtual drive over to a new server? Hello! The server is down! You can't copy sheet. Maybe you can pull something from tape backup. Good luck getting that done in any reasonable amount of time, not to mention all the changes that took place between the backup time and the time of the failure.

Here's another thing to think about for you guys. If you're running virtual servers, you more than likely have several running on a single physical box. So if the box goes down, you've lost all of those servers, not just one. So, now you have to not only find another physical box that can handle more load, copy the virtual drives from your backup solution, and get them started... but you have to do that multiplied by whatever number of virtual servers you had running on the failed machine.

Virtual servers are mostly only useful for non-critical servers [such as QA or development] that don't do a lot of processing, and if they go down will only piss people off, but not cost the company thousands/millions of dollars. Unless you're in a very large company that has hundreds or thousands of non-critical servers that can be consolidated -- virtualization is NOT a good idea.

Justin Chandler

I would just like to point out that when done properly, virtualization adds in a layer of safety. Because, assuming you run the same virtual machine software, the operating system sees the same hardware every time, regardless of what the physical hardware actually is. So, as long as you keep backups, if your main server with everything on it goes down, and it's virtualized, you can just pop the VMDK file or whatever into another physical server's VMWare program, and be back up and running in the time it takes to copy the file. This is great if you have a hot spare lying around. Then, once you get your main server back up and running (or replaced, whatever), you can just throw the VMDK file back into that, and be running, again.

Virtualization does more than just consolidate hardware, people. It wouldn't have taken off quite as much as it did if that's what its whole purpose was.

"The infrastructure isn't large enough to warrant virtualization"

There are other reasons to virtualize besides having a large infrastructure. Especially since VMware Server (GSX) and ESXi are free, there's no 'right' situation to implement it. Fact is, virtualization makes your environment easier to support and will save you hours of time as an admin, as well as solving problems that cannot be addressed by throwing hardware at it

sounds like you should read her dummies book

btw your video is on par with the BOFH, freaking awesome

But we all need virtualisation. Go on buy an SVC.