So recently some folks in and around the OpenStack ecosystem got angry when Red Hat said it had no plans to support other vendors’ OpenStack implementations.
OpenStack, in case you hadn’t heard, is an open source set of building blocks for cloud infrastructures – compute, storage etc – which maps roughly to Amazon Web Services (AWS) proprietary stack. Initially founded by NASA and Rackspace, the OpenStack community now includes pretty much every vendor in enterprise IT.
The argument seemed to be that Red Hat as an open source company, should support everyone else’s stacks too.
Ben Kepes, in his Forbes.com post on the story, argued that Red Hat Plays Dirty To Lock Its Customers In – So Much For Caring-Sharing Open Source.
Open source is caring and sharing? Sometimes – but it can also be nasty, brutish and short. Red Hat wasn’t founded on altruism- it was founded on pragmatism, doing the best job of packaging Linux for the enterprise, making it certifiable and above all trusted as a deployment platform for enterprise apps. As such Red Hat supports a range of third party apps – Oracle, SAP etc – that run on Linux, but doesn’t support other Linux distributions per se. Latterly Red Hat has changed positioning somewhat, responding to the rising importance of developers and bottom-up adoption with its embrace of CentOS.
i have written before about OpenStack’s community before: it’s so big and inclusive that it is unwieldy. Different components are evolving at different speeds, and forking is an issue. That combinatorials at Open Stack are hard to manage, and therefore even harder to support.
This statement from Red Hat seems reasonable.
100% of Red Hat’s offerings are open source, meaning users can deploy and run them anywhere they choose. To meet our customers’ stringent mission-critical requirements in a cost-effective manner, and balance the nearly infinite combination of operating systems, hypervisors, cloud platforms, Red Hat fully certifies and supports many specific Red Hat Enterprise Linux footprints on its own and other vendors’ platforms. Where customers have deployed third-party software, drivers and/or uncertified hardware/hypervisors, the longstanding practice of our Global Support Services team has been to work with customers to diagnose the root cause, and when that root cause is the result of an unsupported hardware or software component, we help the customer connect to the provider for support and get back to a working state. Red Hat has sophisticated, proven mechanisms for certifying supportable hardware and software, and we encourage customers to work within supported configurations for optimum customer experience. This is a big part of the subscription value that customers pay Red Hat for, along with automatically distributing patches to Red Hat Enterprise Linux customers. This provides the highest level of assurance that patches work, are secure, and don’t create unwanted side effects. That policy is noted here: http://red.ht/1gwPdJj.
So Red Hat’s move is entirely rational, but it does have significant implications. IBM and HP for example, which both walked Red Hat into their enterprise customers over the years, are both deeply unhappy, giving a shot in the arm to enterprise Ubuntu (which note, now runs on IBM POWER). Mirantis and Canonical have been playing the open card.
Other vendors have been emboldened by Red Hat’s move, but customers are the ones that will decide. It’s not so long ago that Microsoft didn’t support VMware as a production environment. But if customers anoint another OpenStack distro chances are high Red Hat will change its mind.
oh hai disclosure: Canonical, HP, IBM, and Red Hat, are all clients.
No Comments