Response: Midokura Open-Sources MidoNet

Last week Midokura decided to follow Juniper’s lead and make MidoNet an open-source product (after all, a similar move worked really well for Contrail, right?).

I may be too skeptical (again), but I fail to see how this could possibly work.

So far I’ve seen three types of OpenStack users:

  • DIY crowd who will take whatever open-source products that fit their needs, spend however long it takes to make them work together, and never invest a penny in support and services. The really big players (like Amazon, Google or Facebook) are always taking this path, and some large organizations think they can emulate them;
  • Believers in magical cost reduction, who will spend months or years building an open-source OpenStack environment, only to wake up to the real costs of fully supported versions of all products they used to build OpenStack (because their management wouldn’t allow them to run non-supported products in production). VMware just might be cheaper and faster to deploy.
  • Realists who think OpenStack is the way to go, and save time by buying a supposedly-fully-baked and supported distribution from whatever vendor they prefer. Juniper might get some enterprise support fees from these users (at least where they already work with them); I doubt that too many major vendors will include Midonet in their OpenStack distribution.

Have I missed anyone? Are there OpenStack users out there willing to build their own environment from scratch and then pay for support? Document your experiences in the comments!

Finally, Midokura might count on their product getting some mindshare in the DIY crowd thus making the product interesting enough to make Midokura an acquisition target. That will be a tough sell – SDN is bound to start sliding down the slope toward the trough of disillusionment any time now.


  1. "Fail open" isn't just for firewalls any more.
  2. Hi,

    Yes we are migrating our hosting environment from vendor type environment to OpenStack which we built ourselves and will be paying for the support separately. Midokura has been on our list for some time and it is good to see that it is open source now. And yes it is much cheaper than running VMware with all their MSP licensing.
  3. Is controller required at all?
  4. It has a distributed network store where it holds information about the network but it doesn't program the network from a controller. It's similar to Nuage and Contrail in how it works.

    Nuage and Contrail have some additional benefits in how they interop with existing hardware and standards, I don't see much on the Midokura side. The distributed networking stuff in the OS Juno release might make it kind of irrelevant unless they have a nice GUI or some other compelling feature.

    Most of the OS deployments I know of are for testing still. The production ones I know of are from vendors like Mirantis, Ubuntu, Redhat, Cisco, etc. and are turnkey private clouds for the most part. The vast majority are still using ESX since it's so easy to use and manage.
  5. We spent a huge amount of time and resources rolling our own OpenStack environment (running on Cisco UCS hardware). A little under a year into it and we're still working out production issues from time to time. We're fortunate in that our senior openstack guy is brilliant and is able to figure stuff out most of the time, but as we start trying to integrate the network into that environment, it's going to be more and more on me and less and less on him to figure it out.
Add comment