Server Fault is a question and answer site for system and network administrators. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I am planning out a Hyper-V R2 Cluster environment. I was wondering if I went to the effort of deploying one of the few methods to network boot from iSCSI inside the guests, whether the legacy network adapter would switch to a synthetic after boot, or whether the connection could be handed to a synthetic network after boot?

This is obviously for performance reasons. MS suggests that some emulated devices are capable of switching to synthetic with integration services after boot, but doesn't specifically list all which are capable.

share|improve this question
Why would you want to boot from iSCSI inside the guest when you can connect the host/cluster to the iSCSI target and pass it through as a normal drive? This also allows you to restrict the guest from accessing your iSCSI SAN (SAN security is great and all, but "You shall not pass" has a better ring to it). – Chris S Oct 13 '11 at 16:00

The answer to you question is "no." If you boot through the emulated NIC, that's the NIC you'll use with iSCSI for the boot LUN, and performance will be crappy. It would be theoretically possible to construct some sort of shim that booted from one NIC and then logged into the target with the other, but I'm not aware of an existing solution like that.

The technical truth is that Hyper-V has mechanisms for switching from the emulated IDE controller to the synthetic (paravirtualized) path but no such mechanism exists for networking.

share|improve this answer

I don't believe it is possilbe to boot from iSCSI from a Hyper-V VM, and even if you did the performance (via the emualted NIC) probably wouldn't be that good.

In my experience it is common for people to build 'gold' VHDs which are customised and are factory syspreped ready to run Generalise/OOBE on first boot. If you are using SCVMM to manage your virtual environment this method will be familiar.

share|improve this answer
There are a couple of work-arounds that allow you to do it. My iSCSI software does special caching based around each individual server you provision it to, hence the possible willingness to make this happen. There would only be poor performance if the NIC was a legacy NIC still after boot, hence my question; syntehtic NICs don't suffer the same peformance penalty, I believe. – Carl Feb 12 '10 at 5:57

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.