Thanks -- but it's specifically the internal disks that I'm concerned with
-- I'm guessing there is still only one disk controller, which means that I
can't have 2 completely independent I/O domains without some sort of
external boot device (SAN or DAS attached) for the 2nd I/O domain to boot
from.  That's the case with the current T5440s.  I was hoping that the OVM
2.0 feature that allows you to define down to the device (rather than bus)
might be able to help get round that, but it appears (quite resonably) that
one domain still needs to 'own' the bus and if that domain is down nothing
can acess devices on that bus.

---------- Forwarded message ----------
From: LaoTsao 老曹 <[email protected]>
Date: Mon, Oct 4, 2010 at 1:12 PM
Subject: Re: [ldoms-discuss] T3-4 Server, OVM 2.0 and I/O Domains
To: Tony Scully <[email protected]>


hi each T3 chip support 2 x8 pci-e gen2
so 4 T3 will support 8x8 , *assume*  that there will be 4 PCI-E gen2 switch
to provide redundant 2 cpu and 4 cpu configuration
of the 16 x8 PCI-EM gen 2 slots
hth


On 10/4/2010 8:17 AM, Tony Scully wrote:

  Hi,

Does anyone have a detailed bus layout diagram for the new T3-4 servers?
I'm mainly interested in trying to get redunancy by having 2 I/O domains,
but on the current (T5440) servers the only way to have redunant boot disks
for the 2 I/O domains is to use SAN boot or some other external device for
the second I/O domain.  OVM 2.0 has the ability to assign an endpoint device
but it looks from the docs that the domain that has the bus must be up to
allow this access.  I know it's early days for th T3-4 but would apprecaite
any thoughts anyone may have.

Regards,
Tony


_______________________________________________
ldoms-discuss mailing
[email protected]http://mail.opensolaris.org/mailman/listinfo/ldoms-discuss

<<attachment: laotsao.vcf>>

_______________________________________________
ldoms-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/ldoms-discuss

Reply via email to