One of the neat things I like about GNU/Linux is that the desktop is automatically virtual. The X windows system is a networked desktop. It is interesting to see the world catch on to the advantages of desktop virtualization decades after its invention.
Intel has done a survey. Key findings for me:
- “Ninety percent say that desktop virtualization either plays a foundation role or a strong role in their larger IT initiatives.”
- “IT professionals view desktop virtualization as a way of balancing user needs with their own needs for security, management control and cost containment.”
- IT are the most likely users of virtualization according to 78% of respondents with most other users being considered by ~50-60% of respondents. Engineering and R&D are considered less likely to use virtualization (~30%).
- reductions in IT staffing, energy consumption, hardware, licensing and management combined with increased user productivity are all important factors in the calculation of return on investment.
- virtual hosted desktops are the number one model (~75%) followed closely by virtualized application (~50%) with terminal services, OS image streaming and client side virtual container in the range of ~30-40% of respondents.
- Cost is the major barrier.
- Persuading upper management is moderately difficult.
Clearly, these folks are agonizing over choices and hesitating simply because they are tied into M$. They are mostly looking at high-priced firms like M$ and Citrix to help them get out of the hole that M$ and Citrix dug. If they switched to GNU/Linux the decision would be easy. X over secure networks or X over SSH with insecure networks costs almost nothing to implement and it instantly yields a popular method of virtualization, virtual applications, sessions, etc. Again we see that lock-in is the enemy of the people and IT people in particular. Nowhere in the study is price/performance considered. Using GNU/Linux the performance is actually increased because of caching of commonly used files. That is not available with hosted virtual desktops, except for the user’s own files which might be served from a file server but then there are network delays. Those are not a problem if all the users run on one server.
The chief cost of GNU/Linux virtualization using X is the server which will be needed anyway with these other methods and X is the most efficient so ROI will be maximized. In conversions I did in schools the ROI was huge and immediate. No need to wait years to break even. Changing a set of servers costs much less than a set of thick clients. QED. An organization can likely convert existing thick clients to be thin clients merely by changing the BIOS setup to boot PXE and disconnecting the hard drive, perhaps 15 minutes’ work. A new thin client box can be bought for less than $100. So, for the cost of ~$100 one gets an instant ROI of $200 over a thick client. A server costing a few $thousand can run dozens of clients, perhaps $30 per client. At Easterville, years ago, I put 153 clients on six servers (including file server, four terminal servers and a spare) for $1200 X 5 + $1500 = $7500, $49 per client. Considering only the terminal servers, the ratio would be $39 per client. Cost would be ~$20 per client today thanks to Moore’s Law. That’s an insignificant portion of the cost, yet folks still consider cost the largest barrier next to persuading the boss.
Get with it, World. There is a better way to do IT and that is GNU/Linux.