[openstack-dev] [neutron] New cycle started. What are you up to, folks?
gal.sagie at gmail.com
Fri Oct 2 05:19:47 UTC 2015
This is going to be busy cycle for me, but with many exciting and
Kuryr is starting to gain interest from the community and going to be
demoed and presented in OpenStack
Tokyo, i think all this positive feedback means we are on the right track
and we have a very busy
road map for Mitaka.
1) Kuryr spec in Neutron repository  is accepted by most of the
Magnum team and Kuryr team,
We are planning to demo the first milestone in Tokyo
2) Integration with Magnum and support of the nested containers inside
3) Containerised Neutron plugins using Kolla
4) Finalising the generic VIF binding layer for OVS (reference
implementation) which also can be used by OVN
5) Testing and stability
6) Kubernetes Integration
1) Add tags to Neutron resources spec  (hopefully gets accepted by
2) Port forwarding in Neutron  (still needs work to finalise design
details and spec), this is in my view
a very needed feature which one of its use cases is also to support
Docker port mapping for Kuryr
3) Mentoring program - i would like to present an initiative which will
help both new contributors and
experienced members find each other and be able to delegate tasks
due to mutual interest.
I hope this will help both sides in the long run and will simplify
this process for both sides.
Dragonflow Liberty release is already working as a distributed control
plane across all compute nodes
and implements L2, distributed L3 and distributed DHCP.
It has a pluggable DB layer which already support different backends (etcd,
RethinkDB, OVSDB and RAMCloud)
and the process of integrating new DB framework is very simple and easy.
Dragonflow aims at solving some of the known scale/performance/HA/latency
problems in today's SDN.
1) There are some new people joining and starting to work on the project
which means i will have to do some coaching and helping of new
and always welcome new people that are interested with this project.
2) Pluggable DB layer currently support full pro-activeness (all data is
synced with all nodes)
In the next cycle we plan to support selective pro-activeness
(meaning each node
will only be synced with the relevant data depending on the virtual
and proactive-reactive approach with local cache, which means not
all data is fully distributed
but is queried on demand and cached in the local node.
3) Support missing Neutron API (Provider networks)
4) Extendability - We have some interesting ideas in regards to how
or network functions can control parts of Dragonflow pipeline
without code changes and
implement them in a distributed manner. (using OpenFlow)
5) Smart NICS - leveraging HW offloading and current acceleration
techniques with adjusted
Dragonflow pipeline (we already have some partnerships going on in
6) Distributed SNAT/DNAT
7) Scale/performance/testing - using projects like stackforge/shaker and
on large scale deployment
1) OVN integration with Kuryr
2) L3 in OVN
3) Fault detection and management/monitoring in OVN
4) Testing and picking up on tasks
5) Hopefully trying to share some of the Dragonflow concepts (Pluggable
Sharing information and status in my blog  is very important in my view
and i hope to continue
doing so for all the above topics, hopefully providing more visibility and
help to the community
for the above projects.
On Thu, Oct 1, 2015 at 10:32 PM, Ryan Moats <rmoats at us.ibm.com> wrote:
> (With apologies to the Who)...
> "Meet the new things, same as the old things"
> DVR - let's make it real folks :)
> Performance - I keep turning over rocks and finding things that just don't
> make sense to me...
> I suspect others will come a calling as we go...
> Ryan Moats
> OpenStack Development Mailing List (not for usage questions)
> Unsubscribe: OpenStack-dev-request at lists.openstack.org?subject:unsubscribe
Best Regards ,
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the OpenStack-dev