If you're running with Intel VT enabled in BIOS, and you're running Centos 5 VM inside of Centos 5, and during installation of both OSes you include "virtualization" packages, then there's no noticeable performance hit. The kernels are compiled to eliminate any VM overhead, and it's at least 99% as good, using xen.
If you're running some other OS in the VM, it depends on how the kernel was compiled. It's possible for any OS to perform that well, but without those features enabled, there's a significant performance hit. Off the top of my head, I might say as bad as 30%. My personal experience with virtual machines is this - You're trying to leverage all those CPU cores and ram, but your jobs will become either IO or Network bound, so the cpu's will stay in IO wait instead of working. For compute intensive jobs, I put no more than 4 cores into a machine, with PERC raid controllers and hardware acceleration (read/write cache) enabled, and Gb ethernet. Of course, your experience may be different, depending on the types of jobs you'll be running. Mine are engineers with cadence/synopsys/matlab. One thing I'd like to point out though - Dell PowerEdge specifically - You really should install the OpenManage Managed Node, which allows you to administer your machine on https://machine:1311 but this is not possible with CentOS. Dell partnered with RedHat to develop proprietary code to make that work in genuine RHEL, and it's not possible to do the same on centos. Let's suppose you have a raid array with hotspare, and a disk goes bad, and the hotspare takes over, and you replace the bad disk. The only way you can enable your new disk as a hotspare without rebooting, is to use genuine RHEL and OpenManage. > -----Original Message----- > From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On > Behalf Of Scott R. Ehrlich > Sent: Thursday, June 05, 2008 8:55 PM > To: [email protected] > Subject: [BBLISA] PE2950 and Linux virtualization > > What are people's experiences with CentOS 5.0 64-bit installed on dual > 3 Ghz > quad-core PE2950 systems with 32 GB RAM each, high-performance > computing > (applications that tax both the CPUs and RAM), not currently in a > Beowolf > cluster but could adapt to that, and doing so with VMWare or other > vitualization software vs activity being done directly in the OS? > > How much of a performance hit, or gain (I'd presume hit), does > virtualization > cause an application, resulting in what percentage poorer or better > (I'd > presume poorer) performance vs dealing directly with the OS? > > It would be nice to have a VM perform some work, and if a person's code > or > application breaks, have it take down a VM while keeping a machine up, > and not > affecting other people's work. > > It may also depend on if an application or code is written directly > with/for > the physical cpu/hardware vs more general use (VM). > > Thanks for insights and experiences. > > Scott > > _______________________________________________ > bblisa mailing list > [email protected] > http://www.bblisa.org/mailman/listinfo/bblisa _______________________________________________ bblisa mailing list [email protected] http://www.bblisa.org/mailman/listinfo/bblisa
