Reality Check on Ubuntu’s Enterprise Claims

first_imgTags:#enterprise#news Cognitive Automation is the Immediate Future of… IT + Project Management: A Love Affair Massive Non-Desk Workforce is an Opportunity fo… Related Posts center_img 3 Areas of Your Business that Need Tech Now joe brockmeier 1 Ubuntu founder Mark Shuttleworth has found some encouraging statistics on Ubuntu adoption for public-facing Web sites powered by Ubuntu. Unfortunately, Shuttleworth has taken a single data point and tried to suggest that it’s an indicator that companies are choosing Ubuntu over Red Hat Enterprise Linux (RHEL) for “enterprise computing.” In reality, the stats from W3Techs about Web site usage are not a particularly useful tool for divining what companies are using for “large-scale enterprise workloads.”It’s tempting to cut corners when trying to figure out how companies are using Linux, and how much. It’s damn difficult to figure out how many servers are running Linux and which versions of Linux are in use. That’s because most of the Linux usage, Web servers excepted, is behind the firewall. Couple that with the fact that companies deploy a lot of Linux without having to tell anybody about it, and it’s a real puzzler.You can get some idea by looking at paid Linux usage. Red Hat, SUSE and Canonical can point to the number of paid subscriptions in use, though even that is only a rough estimate of how many organizations are using their enterprise distributions. As Red Hat’s Lars Herrmann noted when I asked him about the topic, many enterprise customers have systems running RHEL without subscriptions.When it comes to unpaid Linux, it’s really anybody’s guess. Looking at Web servers gives some indication of distribution adoption, but it doesn’t necessarily reflect what goes on behind the corporate firewall.On W3techs StatisticsLooking specifically at the stats cited by Shuttleworth, we see that Ubuntu now represents 18.4% of the domains surveyed by W3techs, whereas Red Hat has 12.2%. This is out of the top 1 million Web sites according to Alexa. Note that this is sites and not servers. We don’t rightly know whether the domains in question have one server or 1,000 serving the Web site. We also can assume that the top million sites queried include a lot of non-enterprise sites.So this is a faulty picture to begin with, even if one was to assume that Web servers were representative of enterprise workloads.How many servers are in use for each domain? How many of the domains are being hosted on the same infrastructure by a hosting company, while the organization in question has a completely or partially separate infrastructure? We have no idea.But it gets worse, because Shuttleworth omits all other distributions in his original post. CentOS, which (like Ubuntu) is freely distributed, has 29.9% of the sites surveyed. Debian, which Ubuntu is based on, has 30%. Shuttleworth, in the comments, has the chutzpah to claim Debian as part of the “Ubuntu ecosystem” which seems to be ignoring the way the Ubuntu/Debian relationship actually works.But, it’s not at all surprising that Ubuntu has surpassed RHEL for Web site hosting. Many companies choose Ubuntu Long Term Support (LTS) releases for hosting because it has a reasonable and predictable support lifecycle and doesn’t cost anything to deploy. Herrmann acknowledged that a lot of hosting companies prefer other distributions over RHEL for Web hosting because the RHEL subscription price model makes RHEL less attractive for Web hosting. Most of the time, customers don’t need or want to pay for support to run a Web site on Linux.But Shuttleworth’s claim that companies have “started adopting Ubuntu over RHEL for large-scale enterprise workloads, in droves” seems an overly optimistic reading of a single data point that does not actually represent what most would consider “large scale enterprise workloads.”But for actual enterprise workloads, it’s a different story. What we see is that Red Hat is consistently growing its subscription revenues. In the last quarter, Red Hat had $246.5 million in subscription revenues, up 24% year-over-year. That gives some picture into real adoption of RHEL, but Ubuntu? Well, they don’t report their results and consistently decline to give any real details about their success (or lack thereof) in gaining subscribers.But we can reasonably conclude that Ubuntu is not the distribution of choice for enterprise workloads, simply because so little enterprise software is certified to run on Ubuntu.Popular Workloads for UbuntuYou don’t have to take my word for it when I say that Ubuntu is not the distribution of choice for traditional enterprise workloads. All you have to do is read through Canonical’s own server edition survey (PDF), which collected 5,500 responses from Ubuntu users.Ubuntu shines as a Web server, file server, mail server, etc. But these are not what you’d call enterprise workloads. Not exclusively, at least – enterprises may well be using Ubuntu for Web hosting, file serving, etc., but as the survey says on page 4, “when it comes to application usage, the lower scores regarding CRM and ERP reflect the patchy support for Ubuntu among the big vendors in these fields…these figures confirm that Ubuntu remains a bigger player in the infrastructure realm than it is in applications.”When it comes to application and hardware certifications that enterprises care about, Canonical is far behind RHEL and SUSE Linux Enterprise Server.It seems entirely likely that Ubuntu is seeing increased adoption in the enterprise. Given that Ubuntu has now been around long enough for at least some enterprise IT departments to start taking it seriously, and that its numbers in the enterprise have been historically minuscule. To put it another way, Ubuntu has nowhere to go but up in the enterprise.Stephen O’Grady, of RedMonk, said that while they have no good quantitative data to judge enterprise OS adoption on. However, he did say that there’s reason to believe Ubuntu is making some progress. “I think there are reasons to believe that Ubuntu has made real progress from an enterprise adoption perspective. HP’s inclusion of Ubuntu in its enterprise focused public cloud efforts, for example, would have been unthinkable a few years ago.”But Shuttleworth’s claim that companies have “started adopting Ubuntu over RHEL for large-scale enterprise workloads, in droves” seems an overly optimistic reading of a single data point that does not actually represent what most would consider “large scale enterprise workloads.”Maybe Ubuntu will surpass RHEL at some point, but the evidence so far suggests that Canonical is still in third place behind SUSE and RHEL when it comes to real enterprise workloads.last_img