How important is virtualization to commercial UNIX customers?

The answer is 'not very' according to an annual survey performed by the Gabriel Consulting Group titled, "Is Commercial Unix Relevant in the Midmarket." In fact, virtualization capability ranks 9th out of nine deciding factors when selecting commercial UNIX in the survey. Sure, attributes such as availability, stability, operating system quality, predictable performance, vendor support, ease of administration, price, familiarity are all important but ninth is pretty far down the list. But, I think I know why it ranks as low as it does. The answer is that virtualization just isn't one of the required capabilities on commercial UNIX.

It also has to do with economics. You wouldn't purchase a big iron system that you normally run your mission-critical applications and services on to also use as a virtual host. That is unless you're using virtual zones or jails to better leverage your hardware resources and to isolate applications from one another.

Besides, x86 virtualization is inexpensive compared to commercial UNIX implementations. You wouldn't waste your UNIX resources running a Windows web servers--or at least I hope you wouldn't. You could build an equally available x86 cluster for a fraction of the price.

Now, this information comes from the midmarket-sized businesses (4,000 employees and fewer) surveyed. Large businesses place virtualization in the fifth slot overall. That's probably because they're using zones or other vendor-related virtualization solutions.

Unlike smaller companies, Enterprises can better afford to purchase the equipment and the infrastructure to support large-scale virtualization on commercial UNIX.

You'd think that price would affect how companies purchase solutions but it doesn't. Companies choose commercial UNIX not because of it's low price but because of its rock solid stability. Price, in this survey, ranks 7th out of nine.

So, now the question is, "Should commercial UNIX vendors support virtualization?" Is it a waste of time and resources to develop and innovate in an area where there is no clear desire? I think that UNIX vendors should focus on the strengths of the operating system and continue to offer zones as a virtualization solution.

There's no reason to spend profits on further virtualization research. Zones, containers, jails or whatever you want to call them are a mature technology. It works. You can't install Windows in a UNIX container but you can setup multiple "virtual machines" on a single system to, as I said earlier, better leverage and isolate applications.

Zones are an excellent method of handling multiple applications (especially ones that don't play well with others) on a single system. They're easy to configure, they're lightweight, they're secure, they're isolated and they're free: The capability comes standard with the operating system.

Commercial UNIX vendors should provide support for those companies who wish to use virtualization on the big iron but it certainly shouldn't be a primary focus or selling point--because it isn't for their customers.

By the way, for those of you who think that commercial flavors of UNIX are on their way to the scrap heap, think again. Over 80% of the midmarket respondents in the survey report that their commercial UNIX usage will increase or stay the same. So, commercial UNIX is still relevant but virtualization on that platform, not so much.

As an informal poll, how does the commercial UNIX usage look at your company? Is it increasing, decreasing or staying the same?