In general, I find the draft way too vague (particularly sections 6 & 7, so I 
made no comments on them). I know state migration is a very difficult problem, 
so maybe you should limit yourself to a smaller and manageable subset, for 
example VM migration.

Major remarks
=============
4.1 - What is state
===================
* The flow 5-tuple (section 4.1) is the _flow classifier_ (and even the 5-tuple 
would often include context like VRF or VLAN). There are usually _actions_ 
associated with the flow (FW: permit, log ..., LB,NAT: source or destination 
rewrite). It might make sense to import ideas from OpenFlow documents.

* Middlebox state could be associated with interface, but also with VLAN, VRF 
... Interface is just an example of the context I've mentioned in the previous 
bullet.

* I would love to see an example of middleboxes preserving flow state across 
HTTP-in-TCP sessions as an efficiency mechanism. Middleboxes might retain the 
flow state after TCP session termination to process residual packets (in case 
RST or FIN got lost after traversing the middlebox).

4.2 - State versus Policy
=========================
I found the whole section confusing. I always thought the policies were (in 
general) hierarchical set of classifiers and actions, and when a flow state is 
created, a list of policy actions gets attached to the flow.

4.3 - Instantiating state
=========================
I don't understand what the mechanisms for policy creation (and consumer 
equipment in particular) have to do with this draft. Keep it simple and state 
that the mechanisms used to create policies are out-of-scope.

Also, there's no "hybrid mechanism". Either the middlebox supports a state 
signaling protocol (ex: RSVP) or not. STUN is just a necessary kludge trying to 
guess properties of adjacent NAT devices and using them to its advantage.

5.1 - Addresses of moving endpoints
===================================
There are only two major scenarios: either the network address stays unchanged 
or it doesn't. If it does, we're done, if it doesn't then either the endpoint 
support IP(v6)? Mobility or not. If not, all the sessions are lost and we're 
done. Please keep the wording precise.

5.2.1 - VM migration
====================
Please remove all the marketing & hype. Running VM is moved, its IP addresses 
and sessions are preserved. End of story.

5.2.3 - Failover & HA
=====================
Every single VM failover/HA solution I've seen (apart from VMware's FT) 
restarts the service or VM after the failure, losing all sessions (and all the 
state this draft is discussing). Maybe you should mention that instead of more 
marketing fluff.

7.1 - Recognizing when an endpoint has moved
============================================
There are no "unplanned" moves. Even after a failure, a software component 
(cluster manager, HA manager ...) must restart the failed services/VMs.

Minor details:
==============

1 - Introduction
================
"When an endpoint changes its point of attachment to a network, it retains its 
IP address, and the standard 5-tuple used to describe a flow ..."

That is true in VM mobility scenario, but not in a generic case. How about

"When an endpoint changes its point of attachment to a network while retaining 
its IP address, the standard 5-tuple used to describe a flow ..."

You might also consider slightly rewording the definition of "move" in 
Terminology

> -----Original Message-----
> From: [email protected] [mailto:[email protected]] On Behalf Of
> Melinda Shore
> Sent: Friday, August 31, 2012 7:58 PM
> To: [email protected]; Yingjie Gu(yingjie)
> Subject: [nvo3] Fwd: New Version Notification for draft-gu-statemigration-
> framework-02.txt
> 
> We've published a revision of our draft providing a framework for
> migrating flow-associated state between middleboxes.  One use case is VM
> migration.
> 
> Melinda
> 
> 
> -------- Original Message --------
> Subject: New Version Notification for
> draft-gu-statemigration-framework-02.txt
> Date: Thu, 30 Aug 2012 21:37:41 -0700
> From: [email protected]
> To: [email protected]
> CC: [email protected], [email protected]
> 
> 
> A new version of I-D, draft-gu-statemigration-framework-02.txt
> has been successfully submitted by Melinda Shore and posted to the
> IETF repository.
> 
> Filename:      draft-gu-statemigration-framework
> Revision:      02
> Title:                 A Framework and Problem Statement for Flow-associated
> Middlebox
> State Migration
> Creation date:         2012-08-27
> WG ID:                 Individual Submission
> Number of pages: 22
> URL:
> http://www.ietf.org/internet-drafts/draft-gu-statemigration-framework-
> 02.txt
> Status:
> http://datatracker.ietf.org/doc/draft-gu-statemigration-framework
> Htmlized:
> http://tools.ietf.org/html/draft-gu-statemigration-framework-02
> Diff:
> http://www.ietf.org/rfcdiff?url2=draft-gu-statemigration-framework-02
> 
> Abstract:
>     This document presents an initial framework and discussion of the
>     problem of transferring middlebox (for example, firewall or NAT)
>     flow-coupled state from one middlebox to another while the flow is
>     still active.  This has most recently come up in the context of
>     virtual machine (VM) migration between hypervisors, but it is a
>     problem that has appeared in other situations, as well.  We present
>     some of the parameters of the problem, define some language for
>     discussing the problem, and begin to identify a path forward for
>     addressing it.
> 
> 
> 
> 
> 
> The IETF Secretariat
> 
> 
> 
> _______________________________________________
> nvo3 mailing list
> [email protected]
> https://www.ietf.org/mailman/listinfo/nvo3

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

Reply via email to