I'm wondering if Open Solaris + xVM will allow me to do the following:

I currently have 2 Solaris 10 servers sucking down electrons that I would
like to consolidate into 2 xVM instances on a Solaris 11 quad core box.

This is what it looks like:

Internet --->A sol 10 FW/Web B>---->C Sol 10 app/mail D---> local LAN

So interface A is connected to the Internet.  The FW/Web box runs
IP Filter for FW + NAT, plus there are web services on that system.
Interface B connects via crossover cable to interface C on another
box running Solaris 10/IPFilter (again) as an app/file/mail server for the
local lan.

What I'd like to do is UFS dump the OS images into xVM images and connect
them up with the same network topology.  I have enough hardware to
put 6 ethernets into the new box.

So my questions are:
1) Can I dedicate an interface to an xVM instance without plumbing
an IP on the host Sol 11 box such that I can run IPFilter in xVM as a FW
and not have that interface used as an attack vector to the host?

I would then dedicate 3 interfaces to xVM instances.  2 interfaces to the
first Solaris 10 instance, 1 for the connection to the Internet, and the
other to a cross over to another port dedicated to the 2nd Solaris
10 instance.  The 4th could then be shared between the 2nd instance
and the host.

Ideally it, if this could be done as above, It would be cool if I could
somehow virtually plumb two virtual interfaces between xVM instances
for high speed communications w/o having to use physical hardware, and
for increased performance, but again, without having IP plumbed on
the host to use as an attack vector.

Any thoughts?

(Yes I know this is somewhat convoluted perhaps, and eventually once
Solaris 11 is officially released I'll consider using zones and IP
instances, but until then I'd like not to have to reconfigure two
highly customized Solaris 10 systems just to save on power...)
 
 
This message posted from opensolaris.org
_______________________________________________
xen-discuss mailing list
[email protected]

Reply via email to