I’m working on a larger post about the OpenStack Summit around API Implementation vs. Specification. You can have a preview of that AND A LOT OF OTHER STUFF (OpenStack, Crowbar, lunch) in this 40 minute interview w/ Michael Cote.
Setting: Dell World
Interview w/ @Cote at the Hilton Hotel Lobby on 6th street in Austin.
I know that Cote’s post does not have a time marker for easy navigation; however, I added them to help guide your navigation in the interview (link for audio) if you want to jump around.
- 0:00 Introductions
-
1:00 OpenStack
- 1:00 Essex Conference – what is it, naming conventions
- 2:45 Diablo is adding projects from incubation (Keystone, Dashboard,Quantum,
- 5:30 OpenStack vs. Amazon – “OpenStack has ambitions.” We see it as a “platform for innovation.”
- 6:30 OpenStack is a competitor for Amazon. It implements the EC2 APIs.
- 7:30 How are people managing the evolving nature?
- 8:20 We’re going to see OpenStack in production for the next release based on what we see in our deal flow.
- 9:00 Every user that comes on adds momentum
- 9:30 Rackspace setting up the OpenStack foundation is a reflection of the speed of adoption
- 11:15 Our message is “we’re doing it, we’re in the field.” We are very hands on
- 11:15 We chose early on to focus on helping deployment to help drive adoption
- 12:00 “Our first test for partners is: Are you contributing back to the community?”
- 12:44 The community told us “if you are participating then you are going to open source.” Our commits for OpenStack are live and in the open on our github.
- 13:40 Why Github? We’re happy with it.
- 14:20 OpenStack is using Gerrit because they have a gated trunk. They are migrated to Github
- 15:20 APIs have been a big topic for OpenStack
- 16:00 Do you track who is forking and following? Yes. We also have a listserv. We are trying to do a better job managing the Crowbar community. We know we need to do a better job.
- 17:30 OpenStack is defined by its Implementation. That’s “an effective way to move the project forward quickly;” however, we’re getting to a point where people want to use alternate implementations.
- 19:20 Implementation vs Specification is like the SOAP vs REST debate
- 20:05 This is something the community needs to wrestle with
- 21:45 Specification would allow the efforts to scale. The more people consume the API, the more people care about how it operates
- 22:30 “Bugs can become the API”
- 23:10 Asia and Europe are very active. We are seeing a ton of activity overseas.
- 1:00 Essex Conference – what is it, naming conventions
-
23:30 Crowbar
- 24:00 Crowbar arose out of our need to deploy cloud software regardless of customer infrastructure
- 24:45 We would show up and the customer needed all this cloud infrastructure. We created Crowbar because we always needed this
- 26:00 We extended Chef because we had to do the initial bring-up including BIOS and RAID
- 26:45 We added a state machine and an orchestration layer
- 27:45 Updating the system is a huge component. Every month you may be upgrading the infrastructure!
- 28:30 In our lab, we build whole clouds multiple times a day
- 29:45 Crowbar is the “cloud unboxer”
- 30:00 We modularized Crowbar with barclamps. Hadoop and OpenStack are a series of barclamps. Over 5 for each
- 31:00 Barclamps are applied as layers. We are using that as a term to define DevOps
- 31:15 We are using Crowbar to help message that we understand DevOps
- 31:45 Soup vs Sandwich analogy – Images are like soup while DevOps is like a sandwich.
- 32:45 If you don’t want something in a 1000 server deployment, DevOps lets you make a small change. Gives you flexibility.
- 33:45 We added Cloud Foundry
- 34:00 We’ve made it so easy with barclamps that partners are coming to us with ideas for barclamps. It’s like “changing the meat for the sandwich.”
- 34:30 Dreamhost Ceph team created a barclamp and was actually running a majority of the Crowbar demos at the OpenStack conference
- 24:00 Crowbar arose out of our need to deploy cloud software regardless of customer infrastructure
-
35:25 What’s the future for Crowbar?
- 35:30 More aspects of the infrastructure as open source
- 35:45 More Hardware
- 36:00 Multiple operating systems at the same time (XenServer, ESX, etc)
- 36:30 Larger scale
- 36:50 More types of infrastructure: storage & network
- 37:40 Scalr shout out
- 38:00 We know we need to collaborate more with our community
- 38:30 The first step is to download it and try. Read my blog and sign up for the list serve
- 39:00 CROWBAR IS NOT DELL SPECIFIC – we are working with people who want to create support for other vendor’s hardware. This benefits Dell.
- 39:40 We don’t pretend that our customers are single vendor
- 35:30 More aspects of the infrastructure as open source
Pingback: Crowbar: Where its been and where its going « Barton's Blog
Pingback: OpenStack: Five Challenges & Conference Observations « Rob Hirschfeld's Blog