Drops mike. Hello… what’s THAT all about?
While I suspect this won’t get the big press at EMC World (likely will go to VxRack, XtremIO 4.0 – all great stuff, no doubt) – to me, this is perhaps the biggest strategic announcement (see what Manuvir Das has to say about it here).
This is not JUST about the ViPR Controller going opensource. This is about a fundamental EMC strategy shift and embrace of open source. Expect this to be the first of much, much more to come.
Back in May of last year, sitting around a telepresence on the west and east coast HQs, the exec team realized that we were not moving fast enough, and that the new world was accelerating at a rate of innovation we needed to match and exceed.
The new world (different words for this – IDC calls it “Platform 3”, Gartner calls it “mode 2” in Bi-Modal IT) is defined by:
- Different scales: millions vs. billions.
- Different capacities: Terabytes/Petabytes vs. Exabytes.
- Different app architectures: monolithic vs. composite microservices
- Different data persistence models: RDBMS “single version of truth” vs. “Polyglot Persistence”
- Different abstraction/deployment models: VMs vs. Containers.
- Different scaling models: scale up vs. scale out.
- Different management models: Pets vs. Cattle.
- Different operational/organizational/cultural models: IT Ops + ITIL vs. DevOps
… But it’s also different in another way. Universally….
… Wins. Pivotal was also learning (making CF a the Linux Foundation project for governance + open sourcing GreenPlum, HAWQ, Gemfire, embracing ODP with Hortonworks), and so was VMware (VIO, Neutron work through NSX, and now Project Photon and Project Lightwave)
Why is open source so important for “Platform 3”?
To learn more about this dynamic of closed source vs. open source, our conclusions – and the exact details on the licensing model, the governance, where to get the bits, etc… READ ON!
I’ve been doing an experiment everywhere I go – let’s play a game :-)
Close your eyes.
Read the next couple words, and then comment with the FIRST word that appears in your mind… Ready?
…
…
…
“OPEN SOURCE”.
…
…
…
Having done this now with thousands of people in many countries and in many different languages and cultures, I would wager (safely) that the first word that appears in your brain is “FREE”.
When I’ve done this – what’s interesting is that when you talk to people who are all wrapped up in the world of “P3” work, code and stacks – that’s NOT what they say.
They say these words:
- “COMMUNITY”
- “FRICTONLESS”
- “EASY”
- “INNOVATION”
- “CONTRIBUTION”
- “NO CONTROL = NO LOCK-IN”
This motivation is most true of IP stacks that are “glue” (because they become a lock-in / control point). That’s why we’re doing the ViPR Controller first, because it’s “glue”. It seems less critical (but still important!) for things that are data plane things.
A few months later (when we already had decided we were going down this path), we brought our largest ViPR Controller customer in for a council on the topic – because we wanted them to guide us a bit. ViPR Controller is successful – we’re at about 1000 customers, and nearly universally, people tell us the idea of an open storage abstraction/automation layer is right and necessarily. Invariably – there are things we can do better, and they told us :-) They wanted faster features around more complex layered stacks. They wanted more northbound integration with things like Cisco UCS Director, Service Mesh and Service Now – along with many others – Puppet, Ansible, etc. They wanted more southbound integration with a ton of other storage targets. They wanted better support for brownfield (importing existing storage configs vs. using the ViPR controller to manage net-new storage use).
The biggest single request was: “This idea is now a critical part of our stack – you need to Opensource it. You need to do this for 3 reasons: 1) so we can contribute; 2) so the community can contribute; 3) so that it can never be a control/lock-in point”
When you listen, the voice of the customer is almost always right. In this case, it was the confirmation we were going down the right path.
So – here we are. Here’s the scoop:
- The open source project is called CoprHD (“Copperhead”). Over the last few months, the ViPR Controller team has been shifting their development model to use CoprHD as the trunk – and that will be the going forward model: ViPR Controller is the “EMC Distribution” of CoprHD.
- The github repo with the CoprHD code and CI/CD tools will be up within 30 days. As soon as the repo is up, I will update this post with the link. It will also be where we are putting all our EMC open-source projects – which are here: EMC{code}
- CoprHD is licensed under the Mozilla Public License (MPL) 2.0 model. This highlights that we are dead serious. Technically, you could see other distributions (with all the poop-flinging, can you imagine a “Pure CoprHD distribution? :-) You could see it being used in other products – but any use must also be licensed under MPL.
- ViPR Controller will periodically branch off the CoprHD trunk, and with some additional features (things like deep authentication, EMC Secure Remote Services connectivity and more), will be the commercial offer from EMC. There’s no change to the economic model for ViPR – there is a software license and a maintenance/support element. Customer feedback from the council and NDA dialog is that the fact that it is an open source project makes ViPR more valuable, but we aren’t increasing the price.
- There wasn’t a natural existing project to contribute CoprHD to. OpenStack I suppose would be possible (ViPR Controller is a variant of the same idea behind Cinder). We contribute directly to Cinder as EMC, and like many things in Open stack, Cinder is pretty linked to other parts of the OpenStack framework, and that’s the direction things seem to want to stay. Of course, ViPR and CoprHD are open – supporting a ton of different northbound frameworks – not just OpenStack and consumers are much more than just Nova. We have build a governance model for people who want to contribute code into CoprHD.
- We have an ecosystem with us here at launch. It includes Intel, SAP and others. The success of this open source project is measured by the number of companies (including our competitors) that join the ecosystem.
Now, this is our first big foray down this open source path. We’re in – but there will be a lot to learn. We’ll take every bit of learning and push it back into CoprHD and other open source efforts. BTW - one of drivers (but not the only driver – the other driver was to help us build Project Caspian – see more on that here) behind the Cloudscaling acquisition was to inject some more experience and leadership in this open source domain into the company.
The ViPR team is now learning about how to use this model to iterate and innovate faster.
… And while yes, we will learn through the CoprHD effort – what we already know is that this is the “first of many”.
There will be more to come – and we have an incredible base of great IP – not just in storage/persistence, but in places that will fascinate (see my OnRack (tm) post).
Would love your feedback – the right move? What would you like to see us open source – and why?
Comments