How much is virtualization used today?

I'll just say that the additional convenience of switching to virtualization has outweighed any performance issues that might have otherwise impacted my opinion.

My stance is that my machines that need raw performance don't get virtualized. Everything else is making the transition as it becomes possible for me to convert them.

I love virtualization, both from the perspective of an administrator, and from the perspective of a developer.

On the admin side you gain a lot of flexibility, and a lot of redundancy. If the VM is too slow, you can easily migrate it to a faster machine, give it more resources. Upgrading is a snap. In the event of a disaster, you can flip the whole thing to a new machine.

On the developer side, you can customize the exact os/software layout you need, and the admins don't give you as much crap since it doesn't eat up a whole machine.

Performance can be an issue, but only if you're either trying to virtualize something that's too resource-intensive to easily virtualize, or if you're overloading a VM server. Mostly people don't even notice.

I know that I have had developers dislike the notion of VMs claiming major performance hits.

You may only know about the Developers because they were the low-hanging fruit.

There are many applications where performance is not adequate. Try telling someone with with 200 concurrent users per Citrix server, how would they like to virtualize? Sure, there are a few case studies where it works for lightly used, well-behaved applications. All it takes is one IE process to go to 100% utilization to take out that single cpu guest.

The same performance barrier exists with the Exchange server hosting 5000+ mailboxes or a SQL server that processes hundreds or thousands of queries per second. How much will they benefit from virtualization? Probably none.

Also consider costs. How much more expensive is it to virtualize one physical into multiple guests, to achieve the same performance? The additional Windows os licenses and any software that is licensed per-server, the costs can be prohibitive. Increased administrative effort due to more servers?

Server virtualization was originally intended to consolidate servers that are lightly used or have copious amounts of idle time. That is a good strategy. If someone wants to pursue a broader strategy, they had better make sure it works. Measurements like "how many servers can we virtualize" ring false. The objectives should be finding a good fit and reducing costs.

From my view of it. Its all about doing it correctly. If you do not spend your money where needed like on a SAN or hard drives in general you get what you pay for. Its expensive but in the long run there are a lot of added benefits depending on the budget and what you expect out of the solution. I have not heard of performance problems when the right hardware per need is used. I have been involved with virtual server technology since 2006. Every company I have worked for I have implemented virtual technology and have not heard problems stemming form those solutions put in place.

developers dislike the notion of VMs claiming major performance hits

Developer rule number 1 - "it's not my fault it must be the sysadmins" Of course developers think virtualization hurts their app. Numbers show otherwise. I'm a big fan of virtualizationand encourage all of my customers to consider how they can save time, money and spac doing so. Both Microsoft and VMware have healthy offerings so you can choose a vendor that fits your needs.

I think that it is much easier to implement and maintain than physical servers.

Maintanance should be very similar if not the same, but deployment is much easier (which may not be a great thing). Unless you are deploying servers alot you are not going to see much saving there, however Licensing costs can often be less expensive