[openstack-dev] [FaaS] Introduce a FaaS project

Lingxian Kong anlin.kong at gmail.com
Mon May 15 01:36:14 UTC 2017

Yes, I am recreating the wheels :-)

I am sending this email not intend to say Qinling[1] project is a better
option than others as a project of function as a service, I just provide
possibility for developers/operators already in OpenStack world, and try my
luck to seek people who have the same interest in serverless area and
together to make it more and more mature if possible, because I see
becomes more and more popular in current agile IT world but I don't see
is a good candidate in OpenStack ecosystem.

I remember I asked the question that if we have a FaaS available project in
OpenStack, what I got are something like: Picasso[2], OpenWhisk[3], etc, but
IMHO, none of them can be well integrated with OpenStack ecosystem. I don't
mean they are not good, on the contrary, they are good, especially OpenWhisk
which is already deployed and available in IBM Bluemix production. Picasso
only a very thin proxy layer to IronFunctions which is an open source
comes from Iron.io company who also has a commercial FaaS product.

However, there are several reasons make me create a new project:

- Maybe not many OpenStack operators/developers want to touch a project
  written in another programming language besides Python (and maybe Go? not
  the result of TC resolution). The deployment/dependency management/code
  maintenance will bring much more overhead.

- I'd like to see a project which is using the similar
  components/infrastructure as most of the other OpenStack projects, e.g.
  keystone authentication, message queue(in order to receive notification
  Panko then trigger functions), database, oslo library, swift(for code
  package storage), etc. Of course, I could directly contribute and modify
  some existing project(e.g. Picasso) to satisfy these conditions, but I am
  afraid the time and effort it could take is exactly the same as if I
  a new one.

- I'd like to see a project with no vendor/platform lock-in. Most of the
  projects are based on one specific container orchestration platform or
  to promote usage of its own commercial product. For me, it's always a good
  thing to have more technical options when evaluating a new service.

Qinling project is still at the very very early stage. I created it one
month ago
and work on it only in my spare time. But it works, you can see a basic
introduction in README.rst and give it a try. A lot of things are still
missing, CLI, UT, devstack plugin, UI, etc.

Of course, you can ignore me (still appreciate you read here) if you think
it's really not necessary and stupid to create such a project in OpenStack,
or you can join me to discuss what we could do to improve it gradually and
provide a better option for a real function as a service to people in
OpenStack world.

[1]: https://github.com/LingxianKong/qinling
[2]: https://github.com/openstack/picasso
[3]: https://github.com/openwhisk/openwhisk

Lingxian Kong (Larry)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack-dev/attachments/20170515/fa8df2cf/attachment.html>

More information about the OpenStack-dev mailing list