Loop protection on a virtual switch?

[maulermark]

I have a dual-homed Windows 2008 R2 server that we use for some network training classes.  One interface is a management interface used by students to RDP to the system.  The other interface is a "Lab" interface used to attach to various network devices (switches, routers, etc...) through a layer one switch.  Each interface is on its own separate virtual switch.

The student follows a lab guide and works through various network configuration steps.  Occasionally the students will have a misconfigured switch or router that causes a loop or broadcast storm on the Lab interface.  This seems to take down the management interface also and makes RDP fail.  The only way to get RDP access back over the management network is to use vCenter and disconnect the Lab network interface from the layer one switch via Edit Settings.

The lab network (10.8.10.10) is the default network including the default gateway.  We add a static route for the management network (93.63.1.48) with a metric of 100.  See attachment.

I need a way to protect the management network from the traffic flowing over the Lab network on the system.

The only other option I see that may work is traffic shaping on the virtual switch.

Any other suggestions?

-Mark

We're on ESX 4.1.0 build 320092

[mittim12]

Welcome to the forums.  I have moved this over to the networking forum in hopes that you will get more exposure.

[rickardnobel]

This Windows 2008 R2, is that a virtual machine? If so, how are the virtual network setup at the moment?

[maulermark]

We have a separate vSwitch for each network (Management and Lab) on the server.  The management vSwitch is shared with other servers.  The Lab vSwitch is exclusive to the device.  We have promiscuous mode turned on on the Lab network vSwitch.

I've attached a screenshot.

To recap - my problem is that excessive broadcast or loop traffic from the Lab network knocks out connectivity on the managment network.  Students are unable to RDP to the server through the management network.  Only a hard disconnect of the Lab network and a reboot allows the server to be accessible again.

I'm looking for some way to shield the management network from the loop traffic on the Lab network.

-Mark