Request for virtualization information
Author |
Message |
btiffin
|
Posted: Sat Apr 11, 2009 12:00 pm Post subject: Request for virtualization information |
|
|
Hello gentle folk,
I need to get my head around a potential virtualization project and I'm following the AAFH (ask a friendly human) before RTFM.
Pound for pound and Hz for Hz, how many chips/machines can a single beefy machine emulate before statistical analysis becomes useless? I assume there can be some flexibility in division of the numbers? Assuming (just for instance numbers) a .8 GHz clock and .4GHz bus and an expectation of 100% usage say 5% of the time, is the math of virtualization straight up or some fairly complex formula? Can a 1.6 Ghz dual core machine pretend to be upwards of 80 virtual boxes (given the assumed .8 with max out of 1/20th of the time) and still get meaningful load performance evaluations? What percentage is lost to virtual machine control? Etc...
The goal is to maximize the resources in the lab (the rose coloured beer goggles), while still remaining somewhat realistic in reporting results and avoiding disappointment in the field (the reality of tri-focals).
Just curious, and wondering if anyone here has any technical deets, opinions, experience. And yes, the next step is RTFM, but I prefer AAF compsci.ca H
Cheers |
|
|
|
|
|
Sponsor Sponsor
|
|
|
md
|
Posted: Sun Apr 12, 2009 12:37 pm Post subject: RE:Request for virtualization information |
|
|
Depending on what you are emulating, you can emulate between 0 and an infinite number of machines
It really depends on what specifically you are emulating, and what is being done inside that emulation. For things like running windows in a virtual machine, most computers these days could run a couple instances of 2K, but probably only one of XP or Vista. If your looking to run linux inside linux... well then you could do upwards of 20 or 30 on a single server (actual server...) using some software. |
|
|
|
|
|
|
|