Compression and Cliffs

I set up a number of Windows XP SP2 Virtual PC base images today. A WinXP SP2 clean install, after visiting Windows Update, is 1.70 gigabytes. Building up a few baseline images like this can chew up a substantial amount of disk space and network bandwidth. So, taking a page from Jon Galloway's book, I decided to see what I'd get if I compressed the virtual hard drive file. My results?


This is a companion discussion topic for the original blog entry at: http://www.codinghorror.com/blog/2005/06/compression-and-cliffs.html

Your right Jeff, for example my entire audio library, appx 250 gigs is encoded in FLAC (Free Lossless Audio Codec) which is entirely lossless, and it saves a lot of space. About 33% for FLAC from my experience.

Right, for anyone who hasn’t tried this already, try zipping (or rar-ing) some MP3s for a fun surprise. :wink:

Have you tried the differencing option in the virtual disk? This allows you to have one baseline install, then add SP2, other IE vers etc, but it only stores the data that is different. Works very well when you need a XP, XP SP2, XP w/ misc apps, etc.

Differencing disks – thumbs down, at least with VPC. See here:

http://geekswithblogs.net/jwatson/archive/2005/05/09/39274.aspx

I can’t speak for VMWare, etc.

I guess it is a YMMV. I’ve never had that issue, but mine are very basic and I DO NOT store anything after the baseline is set. For instance, I create baseline XP, then create work XP, XP SP1, XP SP2 differences and install software on them and make them non-persistant from that point forward. In that case it works pretty good, but if you are going to allowing persistance, I can see how the differenced disk could get out of hand.

I have a 300gb drive in my work machine, so space isn’t technically an issue.

It’s more an issue when sharing the GINORMOUS images with other developers, which means copying it over the network and/or writing it to DVD. And that’s where compression comes in handy.

It’s too bad VPC doesn’t offer compressed VHDs an option, actually-- according to the “Virtual PC Guy” disk perf is the #1 bottleneck in VPC right now, but they’ve got non-emulated x86 CPU cycles to burn.

Have a look at this site, here almost all known the compressors are sorted on efficiency :

http://www.maximumcompression.com/data/summary_mf2.php