[openstack-dev] [Neutron] Flavor framework
mestery at mestery.com
Mon Jun 8 13:35:11 UTC 2015
On Mon, Jun 8, 2015 at 8:19 AM, Paul Carver <pcarver at paulcarver.us> wrote:
> What is the status of the flavor framework? Is this the right spec?
Yes, that's the correct spec. The status is that we've committed to
delivering Flavors in Liberty. Doug Wiegley is leading this work along with
some other folks at the moment. I'm hoping we can land the code in
> I'm trying to sort through how the ML3 proposal
> https://review.openstack.org/#/c/105078/ fits in with requirements for
> high performance (high throughput, high packets per second, low latency)
> needed for workloads like VoIP, video and other NFV functions.
I think fundamentally flavors is the right approach for exposing ML3-like
constraints to users. The existing approach specified in the ML3 spec
exposes too many underlying details to users. Flavors can be used to expose
the same thing by operators but without exposing the underlying semantics
of the backend implementation.
> It seems to be the consensus that a significant part of the intent of ML3
> is "flavors for routers" and that makes it natural that ML3 should have a
> dependency on flavor framework but the blueprint referenced above doesn't
> have a series or milestone target.
I've fixed this. :)
Lets see if we can figure out the control plane aspects of ML3 at this
time. I think everyone is in agreement that ML3 makes sense (running
multiple router plugins at the same time), it's just what was exposed to
users which was causing some heartburn. Flavors will address that heartburn.
> OpenStack Development Mailing List (not for usage questions)
> Unsubscribe: OpenStack-dev-request at lists.openstack.org?subject:unsubscribe
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the OpenStack-dev