To Core or Not To Core With Hyper-V

This isn’t really instructions or a recommendation, just me thinking through the keyboard.  I’ve attended Will Craddock’s excellent session on virtualising SBS and EBS and he talks about various strategies.  The normal recommendation from MS for Hyper-V is that you use a Core Installation to conserve disk space/RAM utilisation and to minimise patching and the attack surface.  On the face of it that sounds perfect.  To his credit Will is a realist so he discusses that a little more than most will.

I originally wanted to go with a Core installation but it didn’t work out so well.

Benefits

• The parent partition uses less disk: true.
• The parent uses less RAM – not so sure really.  You’re always recommended to go with 2GB and that’s what I do with a full installation too.
• Less patching: true.
• Smaller attack surface: true.
• Less CPU utilisation: probably true but if you have a dedicated parent partition that you rarely log into, it probably uses very little anyway.

Problems

• OEM’s are still providing hardware management that requires a GUI, e.g. HP network configuration utility, or at worst, require you to learn another scripting language.
• Command line is complex, I can’t delegate host administration to non-advanced administrators.
• Troubleshooting is very hard, e.g. what do you do when the machine goes of the network or won’t authenticate anymore.

I really like the idea of Hyper-V Server 2008.  It solves some of the above by providing a DOS style menu.  R2 also adds CSV and cluster support which is sweet.  I can see that being popular for VDI and labs where OpsMgr and VMM management aren’t required – as of now you cannot manage the free Hyper-V Server 2008 using System Center.

I’m not going to say what’s right for you.  Try out the different solutions and do what suits you best.  The above may be OK for me but you find the Core installation is right for you.