Ubuntu: More Popular Than Red Hat Enterprise Linux?
Mid-March is a time ripe for overhauling traditional power structures — or it was, at least, in ancient Rome. But if former Canonical CEO Mark Shuttleworth is correct in his claim this week that Ubuntu is overtaking Red Hat in the world of Linux servers, we may be in the midst of a major shake-up in the open-source hierarchy as well. Is Shuttleworth right, or will this trend prove no more enduring than Brutus‘ brief reign after he murdered Caesar? Here are some thoughts.
If Ubuntu is overtaking Red Hat Enterprise Linux in the server world, it would represent a major coup. To be sure, Ubuntu is a formidable operating system on servers. I wouldn’t knock it. But RHEL has been around far longer than Ubuntu, and Red Hat, which has traditionally focused nearly all of its commercial operations on the server, is a much larger company than Canonical, which casts a wider net. Red Hat also enjoys entrenched partnerships with a number of major movers-and-shakers in the open-source channel.
In short, Ubuntu’s importance on servers should not be written off. But even Canonical’s own survey of Ubuntu server users, whose results were released last month, seemed to suggest that there were more hobbyists deploying Ubuntu than serious IT operations. To assert now that Ubuntu is more popular than RHEL is thus a big deal.
The Claim
With these stakes in mind, let’s evaluate Shuttleworth’s claim, made on his personal blog, that “A remarkable thing happened this year: companies started adopting Ubuntu over RHEL for large-scale enterprise workloads.” His evidence is this graph, from w3techs.com:
These are interesting statistics, but they’re statistics all the same — which means they have a propensity to distort reality. And indeed it’s not at all clear to me that these figures back up the claim Shuttleworth made.
First of all, this data doesn’t necessarily mean all of Ubuntu’s growth on servers came at RHEL’s expense, which is what Shuttleworth implied when he wrote that users “started adopting Ubuntu over RHEL.”
Second, it’s not clear from these numbers that companies in particular are using Ubuntu more than RHEL, as Shuttleworth also claimed. The figures on the graph represent public websites as a whole, not just those run by companies.
Third, the graph compares the popularity of Ubuntu and RHEL on public Web servers, not “large-scale enterprise workloads” as a whole. Web servers are a substantial part of that equation, but the data I’d really love to see would be a comparison of the two platforms in public and private clouds, which is probably the most important growth area for open-source servers at the moment.
And last but not least, Ubuntu’s expansion is less impressive when one considers market share among Linux distributions overall. In that group, both Ubuntu and RHEL still lag far behind Debian and CentOS.
Ubuntu’s Future
Objectively speaking, the fact that Ubuntu has surpassed RHEL on servers — or, more specifically, on public Web servers tracked by w3techs.com, whether or not they’re run by companies — is impressive. There is a legitimately important trend here worthy of attention.
There’s also good reason to expect Ubuntu to continue to grow on servers, especially in coming months after the next longterm-support (LTS) release of the operating system. Canonical has been doing something right in the server world, and it’s unlikely to stop now.
But if there’s one thing that the data highlighted by Shuttleworth makes clear after a more conservative interpretation than the one he offered, it’s that Ubuntu’s future in the server world, at least for a long time to come, will remain one of growth, not dominance. Ubuntu still faces stiff competition and an uphill battle against competitors in the open-source and proprietary worlds alike. The last thing Canonical should do is declare victory and rest — much like Caesar — on its laurels.
I could not find the sample size for those statistics. Do you have that?
One could fairly safely assume that the boxes running RHEL in that survey were all paying customers given the nature of RHEL. Regarding those using Ubuntu Server, how many are paying customers? I think that is the real question to ask. While they do list SuSE, is that OpenSUSE or SLES?
In any event, Ubuntu Server doing well in the narrow market measured by the survey would be good news for them… in a publicity way… but if only a small fraction of those are paying customers (which would be my guess) it doesn’t really say anything positive about Canonical’s march torward profitability.
The fact that Red Hat continues to have a growing bottom line quarter after quarter (without sacraficing their free software / open source principles) is an indicator to me that the growth of Ubuntu Server in this market area hasn’t really had any measurable impact on Red Hat… but maybe that is stretching it without any real data… similiar to Mr. Shuttleworth’s claim.
The fact that Mr. Shuttleworth is trotting this out as proof of something makes me believe he doesn’t have any credible numbers / measurements which is, at this stage of the game, a bit sad.
Joel: according to http://w3techs.com/faq: For the surveys, we count the top 1 million websites according to Alexa. So that seems to be the sample size.
Scott: agreed. I thought it was kind of embarrassing for Canonical that Shuttleworth is touting statistics gathered by a third party. Surely Canonical must have its own numbers and it would be interesting to see them shared publicly.
Scott: “One could fairly safely assume that the boxes running RHEL in that survey were all paying customers given the nature of RHEL. Regarding those using Ubuntu Server, how many are paying customers? I think that is the real question to ask.”
Very astute. Jef Spaleta and I have asked Mark that question, several times, in the comments on his blog post. It seems difficult to imagine how Mark would not have access to that data. He has not, however, provided it.
Perhaps they used third party statistics to avoid the appearance of skewing data in their favor or in any way omitting/modifying data that would show themselves in a more favorable light. Using someone else’s statistics that are presumably easily verifiable would not lend itself to that point of view.
Since RedHat is in the business of selling services, the number of installations really doesn’t mean much. The amount of services sold does.
This doesn’t mean nothing… In my $dayjob RHEL is extensivelly used, and we have hundreds of RHEL server (I would dare to say that across the entire group, it would escalate to thousands). Less than 1% of those machines are seen from the Internet…
A question for Mark would be… “Paying customers like fancy stuff like Oracle… Is Ubuntu certified for it?” — until there’s a clear answer to that, the only thing Ubuntu might be good at is for children to play 🙂
[…] on the server. But Shuttleworth’s suggestion that Ubuntu is winning at Red Hat’s expense appears to be somewhat flawed. Fast forward to Red Hat’s earnings results and we gain these […]
[…] on the server. But Shuttleworth’s suggestion that Ubuntu is winning at Red Hat’s expense appears to be somewhat flawed. Fast forward to Red Hat’s earnings results and we gain these reminders:1. Real Revenues: Red Hat […]