When implementing vlan taggging the interface becomes partitioned

Bge0 = bge43000, bge44000, bge45000, bge46000

Does this partitioning remove the ability for a service such as
bootparamd to converse over the interface (IPMP Solaris 8 bug and what
appears to be the reason for using the data address on one of the
physical interfaces in an IPMP in all documentation since - see
http://www.sun.com/blueprints/1102/806-7230.pdf)

Will the same IPMP issues exist if we partition the interface with VLAN
tagging?  Or does VLAN Tagging make them improbable as those particular
services may not converse properly over a VLAN tagged interface?

-- Ron

-----Original Message-----
From: Peter Memishian [mailto:[EMAIL PROTECTED] 
Sent: Tuesday, April 25, 2006 4:13 AM
To: Ronald DeMena
Cc: 'Mike Gerdts'; 'Peter Memishian';
[email protected]; [EMAIL PROTECTED]; De Mena, Ron
Subject: RE: [networking-discuss] Official Configuration & Standards
with current IPMP (S8, S9, S10)


 > So does the VLAN Tagging get rid of the "problems" of binding the the
> physical interface?

Sorry, I don't follow.  Could you say more?

 > I have tested the VLAN Tagging using probe based detection in S9, and
it  > works fine with the test addresses on the vlan interface, although
I have  > not come across an app I have tested yet that binds to the
adapter like  > bootparamd does per the blueprints (I don't think
bootparamd would like a  > VLAN interface anyway).  However because of
the way the sun blueprints and  > other sunsolve docs implement the data
address against one of the physical  > interfaces, we are questioning
whether the VLAN Tagging masks the situation  > and makes the clean
configuration with both test addresses on the vlan  > interfaces with
the data on the vlan-sub-interface a viable option.  I also  > want to
move away from active-standby configuration and use it in  >
active-active.
 >
 > Is any of the IPMP re-architecture work done in nevada at this time?

At this point, only very basic functionality is working in my
development workspace.  It will probably go back to Solaris 11 a single
putback, so integration is some time away.  I cannot comment too much on
internal schedules, but rest assured that once we have bits that are
worthy of external experimentation, we will provide early access.

--
meem
_______________________________________________
networking-discuss mailing list
[email protected]

Reply via email to