Red Hat has become the leading vendor of OpenStack, but the company — and others — freely acknowledge serious issues related to complexity, scalability, and availability
OpenStack found its way into four of Red Hat’s top 30 deals last quarter, a quarter that saw the company pushing toward $2 billion in annual revenues — good for Red Hat, good for OpenStack.
Despite that rosy news, however, Red Hat’s earnings call highlighted several areas where OpenStack continues to fall short.
Learn to crunch big data with R Deep Dive promo
Quick guide: Learn to crunch big data with R
Get started using the open source R programming language to do statistical computing and graphics on
[ InfoWorld unveils the Bossies — the best open source products of the year. | Track the latest trends in open source with InfoWorld’s Linux report newsletter. ]
OpenStack still has a long way to go
Take, for example, the OpenStack code base, which continues to be a cauldron of competing projects and a fair amount of poor-quality code.
This isn’t exactly news, given that early OpenStack leader Andrew Shafer once pilloried parts of the project as “a mishmash of naive ideas and pipe dreams.” More recently, OpenStack luminary Randy Bias has candidly derided the silos that different vendors impose on OpenStack, containing “special features that only you have.”
The result? “Every OpenStack deployment is its own unique snowflake,” Bias notes, due to the “hundreds upon hundreds of configuration options.”
Even Red Hat has gotten into the criticism game, with CEO Jim Whitehurst acknowledging OpenStack’s scalability issues on the Red Hat earnings call: “One of the issues … with OpenStack is its scalability, but it’s basically … assuming applications that are stateless,” leaving Red Hat needing to “continu[e] to build more high-availability features to allow it to run traditional applications.”
Despite such deficiencies, OpenStack’s complexity is like red meat for Red Hat, which thrives on taking complex infrastructure and making it easily consumable by mainstream enterprises. As former Red Hat CTO Brian Stevens told me in 2006, “Red Hat’s model works because of the complexity of the technology we work with. An operating platform has a lot of moving parts, and customers are willing to pay to be insulated from that complexity.”
While Bias may not like the vendor-imposed complexity, you won’t hear Red Hat complaining about it. In fact, Red Hat is counting on complexity all the way to the bank.
Enterprises still aren’t thinking about OpenStack correctly
However, OpenStack is still early days for Red Hat customers, Whitehurst revealed during the earnings call. The only ones embracing OpenStack are “earlier adopters,” he suggested, with OpenStack deals “lumpy,” meaning sporadic and large when they do close (due to professional services required to make it work).
Among those early adopters, Red Hat may have a problem. They may may not be ready for what OpenStack was designed: cloud-native applications. As Bias says:
There is no doubt that OpenStack was designed as an AWS clone — that is its lineage. OpenStack is for cloud-native applications… It’s not for running applications that require a 5-9’s infrastructure. [Those] don’t belong on OpenStack.
But Whitehurst, speaking on the Red Hat earnings call, gave a somewhat different view as to Red Hat’s customers and their intentions:
One of the issues or features with OpenStack is its scalability, but it’s basically … assuming applications that are stateless, and so continuing to build more high-availability features to allow it to run traditional applications is something we’ve been talking a lot to customers about.
[T]here’s a general belief that OpenStack is going to be a kind of low-cost platform of choice with customers going forward, but there’s a sense that, hey, Red Hat, you need to help us take some of our existing applications and migrate them onto OpenStack. So we’re actively working with some customers on that.
Such legacy applications aren’t likely to be a good fit for a cloud-native platform.
Of course, the kinds of early adopters interested in OpenStack also have big wallets. While Whitehurst didn’t go into detail, he did highlight the kinds of companies that currently are willing to put up with OpenStack’s rough edges and pay up for professional services to fill in the many blanks it leaves: one of the “very large global telcos” and “a very, very large financial service institution.”
In other words, unless you’re big with serious technology chops, you’re probably not safe going into the OpenStack water. Otherwise, be prepared to pay equally big professional services fees.
But wait! There’s more.
Docker is a much bigger deal than OpenStack
As big as the community behind OpenStack has been, Whitehurst declared Docker the “single biggest topic that comes up among … [Red Hat’s] leading [customers].” In fact, Whitehurst noted that he hears more from customers about Docker than OpenStack.
I’ve argued before that Red Hat should forget OpenStack and double-down on Docker. Listening to the earnings call, this argument gains even more force. Whitehurst talked about why Docker containers are such a big deal. It’s “not because the infrastructure people necessarily want [Docker],” he said, “but [because] developers are picking it up because it’s so much more productive for developers.”
OpenStack is a vendor response to Amazon Web Services — and a half-baked one. Containers, by contrast, immediately make developers’ lives easier (just as AWS does), so they’re being adopted in droves. Developers aren’t asking for OpenStack, and it’s the developer that Red Hat must satisfy.
Let’s review: According to the leading vendor of OpenStack, the technology isn’t mature and as a result is expensive to implement successfully. Enterprises continue looking to OpenStack to solve problems its ill-equipped to solve — and Docker attracts vastly more interest because it meets real developer needs.
Why continue to pour resources into OpenStack?