Virtual Server Hard Drive Configuration


Recommended Posts

We have a few "servers" (really workstation PC's acting as servers) sitting around doing some small tasks. I have a spare server (Older Compaq ML380, 2x 900 Mhz, 2GB RAM, 3x 74GB HDD in RAID 5) and I decided to play around with Virtual Server to get these small sudo servers off of the workstations and onto a real box.

I have setup Virtual Server 2005 (Very Easy) and configured a base Windows 2003 install. My question is how should I set the virtual hard drives up? I have read a little bit about undo disks but I am not sure if that would allow me to use this base image for all the virtual machines and have the virtual machines just create a single undo file for each virtual server.

Should I just copy that base virtual hard drive to a different location and rename it? Also, what is the difference between a virtual SCSI hard drive and a virtual IDE hard drive other then the size limits?

I never personally use "undo disks". I wouldn't use it on a real (non virtualised) server, so I've never seen a reason to use it.

As for the drive differences:

SCSI will be slightly slower in performance than IDE.

More guest operating systems will have drivers built in for the IDE drive, some operating systems may need extra configuration with SCSI ones.

I think IDE allows the use of 4 or 8 disks, while SCSI allows the use of 24/28 disks (may have to confirm that one though).

In the past I have copied Xen images not too sure on VS. If you manage to copy the image, make sure you don't boot it up as the same time as the other image though (hostname, IP conflicts etc.). You should also change the SID. Theres a tool to do it from sysinternals.

You should use SCSI over IDE since the drivers for the virtual SCSI drives are optimised.

This should answer it:

"Contrary to common sense, the performance of our emulated SCSI controller is slower than that of our emulated IDE controller. The reason for this is that the SCSI controller is a lot more complicated to emulate than the IDE controller. Now - this changes once you have Virtual Machine Additions installed. As part of Virtual Machine Additions we install an accelerated SCSI driver. Once this driver is installed the performance of our emulated SCSI controller is significantly faster than our emulated IDE controller."

source

This topic is now closed to further replies.
  • Recently Browsing   0 members

    • No registered users viewing this page.
  • Posts

    • After watching the Apple event earlier this week it is quite the contrast. Apple is going back and tweaking the code to make things more efficient in may areas of MacOS. Windows is boosting your electric build to hide their issues.
    • It is silly there is no simple way to check whether this profile has been activated. CFRs are normal, but trying to even hide the fact if it's on / off seems silly, especially for something so user-facing. Surely Microsoft is "proud" of their engineering efforts on this one and ought to display it somwhere in the GUI.
    • Many Linux distros are not known for excellent battery life, so I'm not sure that is the best example. A more apt example may be Apple, but Apple's CPUs are simply far more efficient than Intel & AMD at single-threaded tasks like these, so "boosting" is not as power-hungry and less heat-inducing. Not to mention Apple will hardly engage P-cores for basic UI tasks; they use a pretty complicated QoS scheme to only activate P-cores for more serious workloads like HTML / JS execution or decompression or application launch. Microsoft is (smartly) doing it for launch, but also for UI tasks, which is the more nonsensical part: why ... do Windows 11's UIs need modern CPUs to boost? It should load so quickly that there's not even time for the CPU to boost.
    • I've not seen any controlled testing and, judging by Microsoft's mentality, within a year, they'll have added so much more bloat, it'll undo any perceptible latency benefit and we'll have boosted the CPU clocks for nothing.
    • It depends: heat soak is a thing. Initially on cold boot-up, the heatsinks & heatpipes are at ambient temp. After heatsinks & heatpipes warm up (through normal usage), they don't immediately cool to ambient temp when the load goes away. So their baseline is higher and the trigger point for fans is much less stress. Add a few more CPU spikes → it's too hot to stay at the same fan RPM → fans get triggered to start up up much sooner / get triggered to ramp much more quickly.
  • Recent Achievements

    • One Year In
      slackerzz earned a badge
      One Year In
    • One Year In
      highriskpaym earned a badge
      One Year In
    • One Month Later
      highriskpaym earned a badge
      One Month Later
    • Week One Done
      highriskpaym earned a badge
      Week One Done
    • Week One Done
      FBSPL earned a badge
      Week One Done
  • Popular Contributors

    1. 1
      +primortal
      501
    2. 2
      PsYcHoKiLLa
      198
    3. 3
      +Edouard
      157
    4. 4
      Steven P.
      84
    5. 5
      ATLien_0
      74
  • Tell a friend

    Love Neowin? Tell a friend!