Well I missed the first keynote this morning, not on purpose, even it was a Cisco one ;-) As Chicago climate was “foggy” I think I didn’t missed anything.
The second I didn’t want to miss was made by Solomon Hykes on Docker (which, as he rightly said, is the word you can’t miss on the Internet nowadays)
His topic was Docker explained through the ground reasons of its creation perspective. It was interesting to see his ability to step back and have a clear look on all the ways people are using his software, identify them through clear use cases, and looking forward on what his community still has to do in order to cover all the use cases he mentionned. I was pretty impressed by his vision, his humble but decided attitude, his optimism. I think our FLOSS ecosystem as clearly a new star here.
And that’s probably among the reasons why the project is so successful. As I wanted to share my my HP colleagues how much I was impressed, I asked to him after the keynote whether he would accept to be one of our TuXTalk speaker for our FLOSS Profession, and to my surprise he accepted right away. Staying as accessible as that is for me another proof we have a new great flagship thinker. I really look forward listening more lenghtly to his thoughts and of course working with docker as such a clever person has for sure created a clever project ! My revelation of the week.
The last Keynote was from Dirk Hohndel from Intel. He stand up instead of the original Intel speaker who had an issue, and didn’t reused his material but used the 20 minutes of the talk to freely talk about two subjects: IoT and the Cloud (that being warned the day before).
He made a pretty funny talk, gathering easily the devs and devops in the room saying that “The Internet was made of things way before marketers get hold of it” or putting emphasis on us as a community rather than on corporations, or ditching the wireless network of the vent (which BTW was flacky indeed). He used that trick to made it easier for him to have adhesion of the audience, which he got. But at the end he passed few messages: one around the need of an Enterprise Group for OpenStack which was created, and another one around Intel promoting a new open standard and open source implementation for discovering and managing devices part of the IoT. More at openinterconnect.org. But don’t expect too much, as there is only 5 companies involved for now. I’m a bit afraid it could become like wimax in the past. But ok, it was an entertaining talk, and rather good due to the lack of preparation.
I then attended a talk from Linda Wang, Red Hat on Docker usage in Enterprise. I was rather disappointed as it remained a high level presentation without too much concrete. I’d have expect more here. The only interesting aspect was the analogy of docker with appartments in a building vs houses for VMs (cgroups being control of electricity, water, …) and the mention of Kubernetes, a container orchestration & management tool from Google.
As I had appreciated Anita Kuno’s talk at LCA this year I then chose to hear her again talking this time about OpenStack Technical Governance.
And while I knew already quite a lot, I leanred some interesting details about the roles and mechanisms around +1, +2, PTL, Technical Comitee, ATC, the election procedure and its Condorcet method. In particular she explained very well the difference between an OpenStack project (git repo) and an OpenStack program (entity recognized by TC, with a PTL, a mission statement)
With some examples around the theoritical definitions, this will become a very good talk people interested in FLOSS governance should listen to. And to stay around OpenStack, after the lunch, I then passed the rest of the afternoon in the HP Helion Workshop, delivered by Mark Dunnett from HP and coordinated by Sisi Chen from HP.
While I knew already quite a lot about the topic, I learned some additional details that I wasn’t aware of, which was the goal for me to attend, as well as to network with my Helion peers !
Mark passed in my opinion a bit too much time on the reminders around OpenStack, especially for the audience around.
He then detailed precisely the differences between HP Helion OpenStack Community and HP Helion OpenStack (why are our marketing guys making it so difficult to just understand stuff by not adding Enterprise to the last one is out of my understanding, and out of the one of many customers I’m interacting with). He thus underlined in the Community edition vs the Enterprise edition the support of KVM vs KVM + ESX (vCenter needed), the 6 weeks release cycle vs Quarterly release e.g. He also talked about the VXLAN support, Icinga addition and ESX proxying in the enterprise version (again my terminology, not HP’s). And our work on the TripleO and Ironic Programs (thanks Anita !), and their usage in all HP Helion OpenStack versions.
He introduced a new component called sirius for the deployment of our storage systems, I ignored (and thus isn’t in the slideset referenced earlier yet). And explained more precisely than my slides how the HA environment is done with ha-proxy and keepalived added in overcloud controler, longside XtraDB for MySQL and RabbitMQ cluster.
And he contrates with the role of the Overcloud management controler which provides in addition some nova, ceilometer and sherpa services (in non-HA mode).
Finally he gave details on the embedded applications provided such as
- the Distributed Virtual Router (DVR) for ovs available to ease east-west traffic between VMs, solving a performance issue and dependency on the network controller, being a SPOF. The DVR will also ease north-south traffic for floating IPs.
- The L2 Gateway which adds mapping between VXLAN and VLAN (which are not able to communicate otherwise) using HP Network switch 59xx
The workshop should have contained a demo which would have made more concrete and real all the concepts seen and show the added value thatHP brings here by making the installation and preconfiguration of all these components just an easy task that every devops or sysadmin can perform to have a quick OpenStack distribution running. However, the demo had an issue and we weren’t able to go very far. Too bad as this is IMO key in such a workshop. Hopefully next version won’t have that issue.
Anyway A very good entry point for understanding our OpenStack based cloud offering, and I look forward working with them to replicate it in EMEA for our customers.