Some people build their own house. A geometrically larger number of people rent.
This has been, to date, the subtext to Red Hat’s cloud narrative. Privately, Red Hat will tell you about the role they’ve played in building large clouds, public and private. Publicly, apart from table stakes efforts such as making RHEL available on EC2 [coverage], their story has been littered with phrases like “Plan => Build => Manage” or “Reference Architectures.” All of which is fine if your target market is those in the market for private home construction. It is less ideal if you intend to service the rental market as well.
Red Hat is Home Depot, then, to Amazon or Microsoft’s landlord. Or at least Red Hat was, prior to the announcements this morning. To explore the news, let’s turn to the Q&A.
Q: Before we begin, do you have anything to disclose?
A: Yes. Red Hat is a RedMonk customer, as are competitors such as Joyent, Microsoft or VMware. Amazon is not a RedMonk customer.
Q: For those that haven’t seen the news, can you tell us what Red Hat announced?
A: Red Hat has announced the non-production availability of a PaaS offering they are calling OpenShift. The URL for the service went live this morning.
Q: Before we get into the specifics, what’s the background here? Where did OpenShift come from?
A: Although their product catalog to date doesn’t necessarily reflect this, Red Hat has been focused on the cloud for some time. A few years ago during a conversation with CTO Brian Stevens, he mentioned in an offhand way a company called Makara, and recommended that we look at them. Which we did [coverage]. Almost a month to the day after we shot that video, Red Hat acquired Makara. Post acquisition, CEO Issac Roth was tasked with building out Red Hat’s PaaS capabilities and given access to resources across the company, including personnel from Red Hat internal IT and the support teams.
OpenShift is the result of those efforts. The Makara code plus some RHEL isolation features, the JBoss EE runtime, and other pieces gives you a top to bottom technology stack.
Q: Taking it apart, what is OpenShift?
A: OpenShift is a PaaS software layer that Red Hat will run and manage on top of third party providers.
Q: Which providers? Who is Red Hat partnering with?
A: Amazon to begin with, with more to follow.
Q: What is the PaaS? Is it framework oriented?
A: No, the PaaS appears to be focused at the runtime level. At launch, OpenShift is supporting Java, obviously, but also PHP, Python and Ruby. The frameworks are intended to run on OpenShift unaltered, and their early results indicate that that’s the case, with Django, Rails, Twisted, etc running without issue. Like VMware’s Cloud Foundry, OpenShift is essentially a container. Unlike Cloud Foundry, it is not yet available as open source.
Q: It’s not open source? Isn’t that unheard of for Red Hat?
A: It’s a bit of a departure, but the company has promised to make the code available. For now, however, they want to observe it in use as a free service to get a better idea of how it needs to be improved or updated.
Q: Have they provided a timeframe for when the code will be available?
A: Not to me they haven’t. Update: following the publication of this piece, I had the chance to speak to CTO Brian Stevens about the open source question. His comment: “we unequivocally and unambiguously commit to [OpenShift] being open sourced.”
Q: The service is free, however?
A: One level is, yes. At launch, there are two offerings: Express (developer oriented) and Flex (operations oriented). The free Express service is roughly analagous to Heroku, just with more languages. You git push your application to OpenShift, it handles the rest: scaling, provisioning, etc. The actual IaaS foundation – Cloudforms – is opaque, however, to Express users. For those who want more granular control over the environment, there’s Flex. It’s got monitoring and configuration features unavailable in the entry level service, and some higher levels of service – it’s not entirely clear what those are at launch – will be paid offerings. A third offering level – Power – will arrive later, and that adds complete control: root access, custom topologies, the whole kit.
Q: How will Red Hat make money off of the service, then?
A: By charging users who consume more than a certain entry level of service money. From the conversations I have had, it does not appear that Red Hat has finalized the exact pricing model yet. That will presumably be announced once it is production ready.
Q: But if I sign up for the free level of OpenShift, which is running on Amazon, Red Hat is paying my Amazon bills for me?
A: For a certain level of service, yes.
Q: What databases are available?
A: At launch, they led with MySQL for relational needs and MongoDB for document database tasks. Their partner page, however, indicates that Couchbase is up now with EnterpriseDB arriving soon.
Q: Put this in context for us: what is OpenShift comparable to?
A: It’s got features in common with a great many offerings, but the closest is probably Microsoft’s Azure. Like Azure, OpenShift is a service managed and run by the vendor, with platform abstractions layered over the underlying infrastructure. Unlike, Azure, however, OpenShift will be open sourced, will be available on multiple hosts, and supports Python and Ruby. Its scope, meanwhile, is broader than Cloud Foundry, Cloud.com, Eucalyptus, and OpenStack.
Q: Why is Red Hat doing this?
A: Big picture, Red Hat cannot afford to be left behind in the cloud market. As a platform player, they must guarantee their relevance to the next generation of enterprise workloads, whether they’re deployed locally or to the cloud. With regard to lockin and the cloud, Red Hat and its customers’ needs are aligned: both are afraid of users being locked into closed platforms.
If you talk to potential cloud adopters, the risk of being locked in is front and center alongside traditional enterprise concerns such as security and compliance. Red Hat, for its part, obviously needs to avoid history repeating itself. Should the cloud market coalesce around a single, closed source platform as happened in the operating system market, Red Hat’s ability to compete there effectively will be limited.
Preventing lockin, then, is important for vendor and customer alike. Which explains many of OpenShift’s design decisions: the ability to choose from multiple hosts, the support for multiple runtimes, and the eventual release of the code as an open source project.
Q: What does this mean for PaaS more broadly?
A: PaaS has gotten substantially more attention from vendors in recent months, beginning this past December with the acquisition of Heroku by Salesforce [coverage] through last month’s Cloud Foundry launch [coverage] and continuing today, with the introduction of OpenShift. Historically, the platform approach has failed to gain significant ground versus its more elemental alternative, IaaS. But this is, in our view, more attributable to design decisions on the part of initial PaaS providers than an inherent and fatal defect of the model [coverage].
As Heroku proved, PaaS platforms built around open models and standardized componentry are able to attract the developer attention that closed models have thus far been unable to. Vendors have made this adjustment, as we expected:
With a multiple year track record of of anemic adoption, PaaS vendors will adapt to customer demand or they will lose ground. Specifically, expect PaaS vendors to borrow from Heroku’s model [coverage], offering platforms assembled from standard or near standard componentry.
If nothing else, then, OpenShift will serve as further validation of the “open” PaaS model given Red Hat’s visibility and market presence.
Q: What does Red Hat need to do to attract even more developers to the platform?
A: Two things. First, support JavaScript. Second, release the code.
Q: Summing it up, what does OpenShift mean for Red Hat?
A: It means that they finally have a single SKU for cloud products. And that can do nothing but help widen their addressable market.
