I finally got round to looking at Mark Shuttleworth's keynote from the OpenStack summit last week.
He clearly had Red Hat in his cross-hairs during the keynote.
As usual lots of hype, exaggeration and some slides can at best be described as disingenuous.
Some of the usual bullshit with Mark again claiming the lead for KVM (what a joke) but here's a new one ....
So Mark is saying that RHEL is legacy - for scale up and Ubuntu is the future for scale out.
Part of this I can agree on - no one (or very few) in the enterprise deploys Ubuntu for Scale up workloads it's not seen as an enterprise grade distribution and doesn't get wide deployment in the enterprise.
But that doesn't mean that RHEL can't play in the scale out space, quite the contrary. I've lost count of the number of hosting providers, telcos, OEMs and end users who've come to Red Hat saying "we've played with OpenStack on Ubuntu, now we want to deploy OpenStack in production and that means running it on RHEL".
Professional ethics prevent me from naming all the companies that Mark used logos for that had this very discussion with Red Hat - that's something we'll deal with with public references at a later date. I wonder how many of them would stand up and give a public reference for Canonical.

Lots of misdirection in those slides - Amazon is there - is Mark suggesting that Amazon runs on Ubuntu/OpenStack or is it just that people can run Ubuntu on Amazon - in which case why not add VMware and VIrtualBox logos?

I've got to wonder who will pay Canonical for OpenStack (or even Ubuntu)
If you can get the bits for free then why pay? What customers want to pay for is support - for someone to fix their issues, help drive new features into the product, deal with 3rd parties for complicated support interactions, certifications etc. How much of that can you get from Canonical?
Let's look at the elements in an OpenStack solution and see how Canonical stacks up. Let's start at the lowest part of the stack Linux itself. 
Canonical's certainly not known for core Linux contributions - they don't rate at all in kernel development[1] or the core userspace libraries - but instead work on their desktop (Unity and Mir). 
Going one step up the stack to virtualization - KVM (which Mark seems to regularly take some credit for Canonical's leadership on). If Canonical are contributing to KVM then it's as close to zero as to be a rounding error.
It's not as if they are contributing fixes instead of features or solid testing, although I thank them for their bugzillas filed against Fedora ...
Then there's OpenStack[2] for which they are certainly not among the top contributors.

Why does this matter? Does it matter how much code they contribute or how many engineers they have?
It matters because when you decide to deploy your OpenStack cloud you pick a partner to help support you - someone who can fix the issues you have, ideally someone who will test it and fix those issues before you encounter them. You need a partner who can drive new features into the platform (whether that's Linux, KVM or OpenStack) who can test it, fix it (proactively) work with 3rd parties on certification, etc.
What I see from Canonical is "compile and ship" and that's not good enough for the enterprise.

Canonical has done a lot of great things for Linux adoption, but every time that guy gets on stage .........

[1] http://lwn.net/Articles/507986/
[2] http://bitergia.com/public/reports/openstack/2013_04_grizzly/
Shared publiclyView activity