Hi, Eric:
Sorry for my missing your reply for quite a long time.
I believe you understand my concern. However when I revisit the section3.7 of 
PS draft, I feel even confusing, it said:
“
   The optimal forwarding problem applies to both outbound and inbound
   traffic.  For outbound traffic, the choice of outbound router
   determines the path of outgoing traffic from the VM, which may be
   sub-optimal after a VM move.  For inbound traffic, the location of
   the VM within the IP subnet for the VM is not visible to the routers
   beyond the virtual network.  Thus, the routing infrastructure will
   have no information as to which of the two externally visible
   gateways leading into the virtual network would be the better choice
   for reaching a particular VM.
”
My point is multi-homing issue or one network endpoint is connecting to one VN 
through multiple first hop network devices
may complicate optimal forwarding problem but seems not the basic problem for 
optimal forwarding.
The basic problem for optima forwarding is the case each VN has one first hop 
network device or egress router.

For inbound traffic, even you only have one first hop network device that 
connect source network endpoint to one source VN,
The location of the source VM within the source IP subnet(i.e.,VN) is also not 
visible to the destination routers in the destination VN
beyond the source VN.

Also it seems you don’t consider the outbound traffic for the case when VM is 
moved from VN1 to VN2 and each VN have only one first hop network device (could 
be a router)
In such case, the traffic can be forwarded/tunnelled by first hop network 
device in VN1 to first hop network device in VN2, doesn’t it? Why this case is 
excluded?
Also in this case, how the destination VM location within destination VN is 
visible to the source first hop network device in VN1 is still a problem, 
Doesn’t it?

For other minors issues, like
1.terminology network endpoint, end system, endpoint, it is better to use 
consistent terminology.
2. I don’t understand the sentence
“
A virtual network may have two routers for
traffic to/from other VNs or external to all VNs
”
What do you mean by virtual network have routers for traffic to external to all 
VNs.
3. Is router a right term? It seems in other place you use first hop network 
device, it doesn’t exclude switch to be used.
I can come up a specific text proposal if you allow for couples of days. Thanks 
for taking my comment into account.

Regards!
-Qin
发件人: [email protected] [mailto:[email protected]] 代表 Eric Gray
发送时间: 2013年3月14日 23:21
收件人: [email protected]
主题: [nvo3] Comments on draft-ietf-nvo3-overlay-problem-statement

Had to forward this as the commenter used an invalid email address.


Sent from my iPad

Begin forwarded message:
From: Mail Delivery Subsystem 
<[email protected]<mailto:[email protected]>>
Date: March 13, 2013, 5:11:01 PM EDT
To: [email protected]<mailto:[email protected]>
Subject: Delivery Status Notification (Failure)
Delivery to the following recipient failed permanently:

    [email protected]<mailto:[email protected]>

Technical details of permanent failure:
DNS Error: Domain name not found

----- Original message -----

DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed;
       d=gmail.com<http://gmail.com>; s=20120113;
       h=x-received:content-type:content-transfer-encoding:subject:from
        :message-id:date:to:mime-version:x-mailer;
       bh=M1TL6c0sHzx2Id1yUCc17V47J+w39+N0Y9+Zce455O8=;
       b=gpxFHXeLBl1i40U1rIhvlZ6RRApxNFKV7h9pJzFzW6Tx3RjHpCSlDqx9hb/LFbb9Ub
        CYjoacI0UGGUVuA4LN/cwakYxLmBb1yLaIqGF7/7/WQ8HcUG4YRhdALlRT+W8zJ2VVco
        LKEHbnOVEussT4CzTiTGXDaQedrDee7pgkdDs0bpyMtScXo7k7jB7IwlCO3pN71FelY5
        ywk1RgP4qFFuDdXOGsrrXS8JqhO91kUr2rbwgCR2/w90rjsOVX1CB90nha67j0PlaA5w
        2EnOtVKsfS88Js25u9PuT5XvZyzvmEJsUK3dgZWIGHKkJJe9opX31xhzO6Z3+ojQV9yg
        q7lQ==
X-Received: by 10.68.143.167 with SMTP id sf7mr48452806pbb.21.1363209061007;
       Wed, 13 Mar 2013 14:11:01 -0700 (PDT)
Return-Path: <[email protected]<mailto:[email protected]>>
Received: from ?IPv6:2001:df8::16:7d23:71af:5434:aa14? 
([2001:df8:0:16:7d23:71af:5434:aa14])
       by mx.google.com<http://mx.google.com> with ESMTPS id 
qd8sm31117738pbc.29.2013.03.13.14.10.58
       (version=TLSv1 cipher=ECDHE-RSA-RC4-SHA bits=128/128);
       Wed, 13 Mar 2013 14:10:59 -0700 (PDT)
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: quoted-printable
Subject: Your comment on draft-ietf-nvo3-overlay-problem-statement-02
From: Eric Gray <[email protected]<mailto:[email protected]>>
Message-Id: 
<[email protected]<mailto:[email protected]>>
Date: Wed, 13 Mar 2013 17:10:59 -0400
To: "[email protected]<mailto:[email protected]>" 
<[email protected]<mailto:[email protected]>>
Mime-Version: 1.0 (1.0)
X-Mailer: iPad Mail (10B146)

Qin,

In reading your comment, it seems that you're both asking what we
mean by the text and answering the question with your next question.

The text you ask about is identifying the case where more than one
router connects a virtual network to an otherwise separate virtual
network.  The optimal forwarding concern is that it is difficult for a
virtual host or server located within the first virtual network to know
which of the multiple routers connecting the two virtual networks is
the optimal choice for connecting to a host or server in the second
virtual network.

This is a concern when there are possibly multiple local forwarding
devices between the first host/server and any of the potential routers
that connect the two virtual networks and/or a similar situation exists
in the second virtual network between any of these candidate routers
and the second host/server.  How is the host supposed to determine
which of the multiple routers in this case provides the optimal path.

This is pretty much the case you identify when you ask how we are
"going to deal with the optimal forwarding between two VMs [that]
belong to different subnet[s]."

It seems that we're at least talking about the same thing.  Can you
make a specific text proposal that would make our text clearer?

--
Eric

You wrote:
I reviewed the change in section 3.7 "optimal forwarding"of NVO3 problem 
statement.
It said:
"
  IP implementations in network endpoints typically do not distinguish
  between multiple routers on the same subnet - there may only be a
  single default gateway in use, and any use of multiple routers
  usually considers all of them to be one-hop away.
"
It seems you talk about one tenant system is multihomed to multiple NVEs in the 
same subnet.
I am not sure how much of this is related to optimal forwarding?
Are you assuming only one NVE is active? or all the NVEs placed in the same 
subnet are active?
or some of NVE placed in the same subnet are active to a set of VMs while the 
other
_______________________________________________
nvo3 mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/nvo3

Reply via email to