Open Stack

There are some discussions on the spec review and Kyle suggested the discussion should be on openstack-dev. And also we want to get more attention on this topic to ensure that ML3 is the right approach.

Problem description:
At this point, only a single L3 router plugin can be exclusively specified by configuration. As the result, only a single type of router can be deployed. Multiple types of L3 routers can't be deployed simultaneously. For example, neutron native L3 router(legacy/ha/dvr) can't coexist with other opensource/vendor routers.

The basic idea that we are proposing is a L3 router plugin version of ML2. Implement a new modular L3 router plugin and port the existing reference L3 plugin (neutron/services/l3_router/l3_router_plugin.py) to the router driver of L3 router plugin. (Agent side doesn't need modification). It's up to each vendor if they will port their L3 router plugin into ML3 router driver. Armando and  Paul made good points in the spec review ( FYI https://review.openstack.org/#/c/105078/).

The information we are seeking is:
1) Whether there is a broader requirement for running multiple types for L3 routers in OpenStack deployment.
2) Whether ML3 is the right approach to achieve so.

Thanks,
Zang, Rui


From: Kyle Mestery [mailto:mestery at mestery.com]
Sent: Tuesday, June 2, 2015 9:17 PM
To: CARVER, PAUL; Maru Newby; Assaf Muller
Cc: Zang, Rui; Isaku Yamahata; YAMAMOTO Takashi; Stephen Wong; IWAMOTO Toshihiro; Bob Melander; mark mcclain; Carl Baldwin; Salvatore Orlando; Gal Sagie; vikram.choudhary; Armando Migliaccio; Robert Kukura; Sukhdev Kapur; Wang, Yalei; vishwanath jayaraman
Subject: Re: Change in openstack/neutron-specs[master]: L3 plugin: modular l3 router plugin

Two items:

1. This seems like a candidate for an RFE following our new RFE process [1].
2. This discussion really should be happening on openstack-dev.
Thanks,
Kyle

[1] https://github.com/openstack/neutron/blob/master/doc/source/policies/blueprints.rst

On Tue, Jun 2, 2015 at 6:27 AM, CARVER, PAUL <pc2929 at att.com<mailto:pc2929 at att.com>> wrote:
Replying directly rather than adding side noise to review.o.o: Yes, I'm going to send the survey to the OpenStack General list and the OpenStack Operators list. I'm not sure if it's actually appropriate to send the survey to the -dev list although I'll certainly post the summary results there. I just wanted to give Armando and anyone else following the ML3 review discussion an opportunity to comment on whether I've asked the right questions before I send the survey out.

For those who didn't see it, the survey link is here: https://www.surveymonkey.com/s/BTMH7V5 and the purpose is to gather data on how people are currently doing L3 in OpenStack and whether a modular (i.e. supporting multiple L3 technologies/vendors per Neutron instance) L3 is something that people see a need for.

Comments are welcome on the wording of the questions and the multiple choice answers I provided. I'll send the survey out to the mailing lists sometime after noon EDT (UTC-4). Any comments I receive before then, I'll happily tweak the wording before sending.

-----Original Message-----
From: Rui Zang (Code Review) [mailto:review at openstack.org<mailto:review at openstack.org>]
Sent: Tuesday, June 02, 2015 02:31
To: Isaku Yamahata
Cc: YAMAMOTO Takashi; Stephen Wong; IWAMOTO Toshihiro; Bob Melander; mark mcclain; Carl Baldwin; Kyle Mestery; Salvatore Orlando; Gal Sagie; vikram.choudhary; Armando Migliaccio; Robert Kukura; Sukhdev Kapur; CARVER, PAUL; yalei wang; vishwanath jayaraman
Subject: Change in openstack/neutron-specs[master]: L3 plugin: modular l3 router plugin

Rui Zang has posted comments on this change.

Change subject: L3 plugin: modular l3 router plugin
......................................................................


Patch Set 10:

(1 comment)

https://review.openstack.org/#/c/105078/10/specs/liberty/ml3-router-plugin.rst
File specs/liberty/ml3-router-plugin.rst:

Line 24: L3 routers can't be deployed simultaneously. For example, neutron native
> That's a really good point, but I'm not sure what the right answer is. Bein
Paul, I assume you are going to send this survey to the mailing list, right?


--
To view, visit https://review.openstack.org/105078
To unsubscribe, visit https://review.openstack.org/settings

Gerrit-MessageType: comment
Gerrit-Change-Id: I1c476156eb9788056e50e48c85c3d9b17e336b38
Gerrit-PatchSet: 10
Gerrit-Project: openstack/neutron-specs
Gerrit-Branch: master
Gerrit-Owner: Isaku Yamahata <yamahata.review at gmail.com<mailto:yamahata.review at gmail.com>>
Gerrit-Reviewer: Armando Migliaccio <armamig at gmail.com<mailto:armamig at gmail.com>>
Gerrit-Reviewer: Bob Melander <bob.melander at gmail.com<mailto:bob.melander at gmail.com>>
Gerrit-Reviewer: Carl Baldwin <carl at ecbaldwin.net<mailto:carl at ecbaldwin.net>>
Gerrit-Reviewer: Gal Sagie <gal.sagie at huawei.com<mailto:gal.sagie at huawei.com>>
Gerrit-Reviewer: IWAMOTO Toshihiro <iwamoto at valinux.co.jp<mailto:iwamoto at valinux.co.jp>>
Gerrit-Reviewer: Isaku Yamahata <yamahata.review at gmail.com<mailto:yamahata.review at gmail.com>>
Gerrit-Reviewer: Jenkins
Gerrit-Reviewer: Kyle Mestery <mestery at mestery.com<mailto:mestery at mestery.com>>
Gerrit-Reviewer: Paul Carver <pcarver at att.com<mailto:pcarver at att.com>>
Gerrit-Reviewer: Robert Kukura <kukura at noironetworks.com<mailto:kukura at noironetworks.com>>
Gerrit-Reviewer: Rui Zang <rui.zang at intel.com<mailto:rui.zang at intel.com>>
Gerrit-Reviewer: Salvatore Orlando <salv.orlando at gmail.com<mailto:salv.orlando at gmail.com>>
Gerrit-Reviewer: Stephen Wong <stephen.kf.wong at gmail.com<mailto:stephen.kf.wong at gmail.com>>
Gerrit-Reviewer: Sukhdev Kapur <sukhdev at aristanetworks.com<mailto:sukhdev at aristanetworks.com>>
Gerrit-Reviewer: YAMAMOTO Takashi <yamamoto at midokura.com<mailto:yamamoto at midokura.com>>
Gerrit-Reviewer: mark mcclain <mark at mcclain.xyz<mailto:mark at mcclain.xyz>>
Gerrit-Reviewer: vikram.choudhary <vikram.choudhary at huawei.com<mailto:vikram.choudhary at huawei.com>>
Gerrit-Reviewer: vishwanath jayaraman <vishwanathj at hotmail.com<mailto:vishwanathj at hotmail.com>>
Gerrit-Reviewer: yalei wang <yalei.wang at intel.com<mailto:yalei.wang at intel.com>>
Gerrit-HasComments: Yes

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack-dev/attachments/20150604/46cd583c/attachment.html>