Open Stack

Hi Everyone,

Thierry asked each PTL to send out a note about project goals in the
Grizzly release cycle.  I wanted to talk not just about new features being
added to the Quantum codebase, but also what I see as helpful directions
for the Quantum team as a whole.  Since most people just want to read about
the features, I put that at the bottom to make you read the team stuff
first :)

I'm happy to hear other people's inputs on the topic of project goals as
well.  Thanks,

Dan

== Quantum Team Direction ==

Broaden Team Leadership:

As the project grows both in terms of contributors and codebase, we need a
more distributed model of team leadership.  As discussed at the summit,
we'll be experimenting with introducing sub-teams for different aspects of
Quantum, with sub-team leads that are charged with helping to drive
features, monitor bugs, and ensure consistency within their area of
expertise.

Rise of the Users:

Quantum has so far been a project primarily driven by vendors, but with the
transition to core project status, we are seeing more Quantum users get
involved in the development of the project, and even taking leadership
roles.   This is a great trend that we want to build on even more in
Grizzly, as it helps keep the focus on "enabling use cases for cloud
tenants" rather than on "expose vendor capability X".

Focus on Enabling Real-world Deployments:

Now that Quantum is core, its important that we focus not only on adding
new capabilities, but also on recognizing the practical issues (bug fixes,
stable releases, documentation, answering deployment questions, improving
usability, upgrade, etc.) that help people be successful using Quantum.  As
a team, we need to make sure that people who focus on these things are
respected and rewarded and that we don't always rush to implement the "next
big thing".

Growing the Core Developer Team:

Fundamental to Quantum's success is a core team who spend significant
cycles working on community projects and reviewing code contributed by
others.   Making sure these developers, and their employers, see value in
being a core dev and helping this team grow over time without sacrificing
developer quality is central to the long-term success of the project.

== New Quantum Capabilities ==

Improved System Test / Gating:

Our unit tests provide good coverage, but getting this in place is key to
handling a growing feature-set and handling interactions across projects.
We will work closely with the CI and Tempest teams on this.

Close Feature Gaps compared to Nova network:

We want to make sure that Quantum can be used in all major use cases
supported by nova-network.  One key gap here is enabling a deployment model
that has similar properties to the nova-network "multi-host" flag, in which
DHCP + NAT services run locally on each compute node.  Additionally, we
want to update features like security groups (see below) and metadata
server to support overlapping IPs, which is a key benefit of Quantum.

Improved Security Groups:

We will deliver security groups as part of the Quantum API, with several
benefits over standard nova security groups, including: ability to apply
security groups per VM interface (not just per instance), ability to filter
both inbound and outbound traffic, ability to filter v4 and v6 traffic,
ability to change group membership after VM boot, etc.  This not only
meets, but exceeds the capabilities of Amazon's newer version of security
groups (called VPC security groups).  Similar to floating-ips, we will
provide a proxy so that nova security group commands can be passed on to
Quantum.

Load-Balancing and other advanced network services:

With Folsom Quantum delivered a base L2/L3 feature-set, so attention is now
turning to more advanced network services.  We are building a generic
framework to enable multiple plugins, each of which offers a service built
on top of the core L2/L3 Quantum service.  The first higher-layer service
we are focusing on is load-balancing, which will support an open source LB
solution (e.g., HAproxy) as well as vendor technologies.  Once the network
services framework has been proven out with load-balancing, we will explore
additional services such as VPNs and advanced firewalls.  Adding richer
advanced network services will be an ongoing effort that extend beyond the
Grizzly release.

Adding Support for New Backend Technologies:

New vendors continue to express significant interest in integrating with
Quantum.  We expect new plugin proposals to add Quantum support for
technologies like Brocade, Floodlight, Hyper-V virtual networking and
more.  We also expect several vendors to contribute drivers for different
load balancing backends, including Citrix, F5, Radware, Riverbed.  Managing
the contributions for these new parties will be a major challenge for the
core team.

Improve Horizon Integration:

Folsom saw the initial integration of basic Quantum features into Horizon,
but in Grizzly we will add more advanced capabilities, including the
management of L3 routers, security groups, floating IPs, and
load-balancing.
The full list of Grizzly blueprints for Quantum is here:
https://blueprints.launchpad.net/quantum/grizzly

-- 
~~~~~~~~~~~~~~~~~~~~~~~~~~~
Dan Wendlandt
Nicira, Inc: www.nicira.com
twitter: danwendlandt
~~~~~~~~~~~~~~~~~~~~~~~~~~~
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack-dev/attachments/20121107/3b47f7c0/attachment.html>