Hi Yizhou

Thanks for your quick responseŠ When thinking about control plane scale, I
personally want to steer clear of any assumptions that an environment
consists of single IP and MAC per NVE<>TSI broadly and then end up without
the CP scale needed to support an environment that connects numerous VNs,
IPs and MACs per VMs.   I do think your suggestion would help.

Ed

From:  Liyizhou <[email protected]>
Date:  Wednesday, July 20, 2016 at 2:13 PM
To:  Edwin Mallette <[email protected]>, "[email protected]"
<[email protected]>
Subject:  RE: draft-ietf-nvo3-hpvr2nve-cp-req-04 Comment -

Hi Ed,
 
Thank you for your review. The sentence you quoted tries to explain the
requirement to disassociate all (meaning more than one) addresses given
owning single address per VM is pretty common. It has no assumption on if
VMs are or are not highly   connected.
 
My feeling is to make the logic clearer, it would be better to put it before
³In the Split-NVE architecture, the external NVE may be able to reach
multiple MAC and IP addresses via a TSI.²
 
Thanks,
Yizhou
 
 

From: nvo3 [mailto:[email protected]] On Behalf Of Edwin Mallette
Sent: Wednesday, July 20, 2016 7:51 PM
To: [email protected]
Subject: [nvo3] draft-ietf-nvo3-hpvr2nve-cp-req-04 Comment -
 

Hi authors, all,

 

I was reviewing this document and I noted an assumption made that (while
technically true, I don¹t really know but it sounds like it might be true :)
might not be a good or safe assumption based on some implementations that
are almost entirely not the case described here:

 

1.2  Target Scenarios ­ last paragraphŠ

"In the majority of cases, a VM will be acting as a simple host that will
have a single TSI and single MAC and IP visible to the external NVE.²

 

I¹m not quite sure why this is pertinent per se.  These requirements should
clearly account for both highly connected VMs (and implementations that are
90+% that case) and the non-highly connected VMs.  Please let me know if I¹m
not quite thinking about this the right way.

 

Cheers!

 

Ed


_______________________________________________
nvo3 mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/nvo3

Reply via email to