There are certain situations I can agree virtualization can be useful and efficient.
Development/ programming is probably the best environment. One can write code and test it without worry of botching the whole system up.
There are lots of people who push, aggressively, for use of virtualization in servers though and that is an area I strongly disagree with. Especially where Linux and OpenSource is concerned.
Now, everyone has their own opinion and is entitled to it. Some folks are just very excited about the perceived potential in virtualization in regard to maximizing resource effieciency.
The state of computer hardware to date, is made to go bad. It is made to be replaced, not repaired.
It is a concept called "planned obsolescence" and was generally made notorious by AT&T after Ma Bell broke up.
Instead of leasing telephones, they created cheap phones, made of cheap parts, designed to fail after a certain amount of usage. This allowed them to make more money in phone sales by selling the first phone, then replacement phones. A repairman didn't come out to fix the phone anymore. You go to the store and buy another.
They designed it that people wouldn't have to wait very long before the phone needed to be replaced.
The same technological snakeoil is sold by computer and electronics makers today.
The hardware that makes up the computers we use, even those we use as servers, is not made to be hardy and reliable. It is made to fail efficiently. Techs don't really spend a lot of time rebuilding hard drives anymore. They replace hardrives. Repair not only costs too much, but requires more time and training than most are willing to invest in it.
The story is the same for power supplies, optical drives, memory sticks, circuit boards, etc...
Nothing is exactly cheap either. Oh sure, market prices rise and fall, based on supply and demand (whether that demand is artificially created or not) but on the whole, hardware prices increase over time rather than decrease.
If you are a company or person with the financial resources available and willing to allocate to hardware, you can afford to keep a stock of replacements parts, even entire replacement computers ready in the wings to slide in as a backup or failover in case the server should have a hardware breakdown. Then, in this situation, virtualization might be answer for you.
If you do not have the financial resources available or designated to maintain such stock in hardware, the idea of keeping all those services on one machine may be tempting, but are you really willing to risk that motherboard or powersupply going down at the worst time? How long can your servers, all of them on the one machine, be down in order to repair/replace it?
An hour? Two hours? 30 minutes even? What is acceptable down time to have not just one or two services down, but several of them?
Have you ever heard the old saying of not putting all your eggs in one basket? Do you keep all of your money in one bank account? Wise people have learned since the stockmarket crash so many years ago that one does not do that if one wants to keep their money.
If you save a bit of money each month by having all those servers on one physical machine, how much will it cost you in lost production and down time when that one machine fails and those many servers go down without the pre-requisite backup hardware in place?
Yes, virtualization can and does have a place in the server world. It has a limited scope and in that can be a well funded and prepared IT departments every dream.
If you are not that well funded and prepared IT dept, you are taking big risks and I hope you never experience that nightmarish hardware crash.
Thursday, July 31, 2008
Subscribe to:
Post Comments (Atom)
0 comments:
Post a Comment