The very different economics of running open-source software
Open-source pioneer Tom Lane talks us through the tricky calculations surrounding enterprise free-to-use software
With its emphasis on the collaborative rather than the competitive, the economics of open-source software are very different. So what do companies need to understand? Computing asked open-source pioneer Tom Lane for his take.
While not exactly unremarked, the change in the software industry over the past decade has certainly been extraordinary. In essence, an entire industry has taken its primary product and started doling it out for free. That's not to say that the proprietary software is dead - far from it. The Oracles, Microsofts and SAPs are still very much alive and remain vastly profitable, but they, like everyone else, are having to adapt to the new way of doing things.
For most enterprise software projects these days, rather than an initial phone call to an Oracle VAR, the first step is checking projects on Github to see if any could be adapted to fit.
Tom Lane knows more about this change than most, having helped to drive it. According to his Wikipedia page, in 2000 0.782 per cent of all open-source code could be attributed to him. Instrumental in developing image formats JPG, TIFF and PNG, Lane also sits on the core committee of the PostgreSQL database project, having been engaged in that project since 1998. He recently took up a position as elite PostgreSQL developer with Crunchy Data, a US supplier of PostgreSQL that adds security, scalability and other enterprise features.
So, as open-source has moved centre stage, is there any reason why an organisation should look at a proprietary solution first any more? Lane says that, broadly speaking, there are still two scenarios where that makes sense.
"One such case is where there's simply not a likelihood that you can get a development community to coalesce around the project, perhaps because it's too small a niche case, or the barrier to entry for development is too high for some reason," Lane explains.
Another, he goes on, is where a company's intellectual property is intimately tied up with the project and so must be kept confidential: "Most of the successful open-source projects I know about are working in fields where the basic techniques have been around for decades, so there's no private IP needed."
However, he says, in most cases it makes sense to look at open-source first.
Costs and benefits
A cost-benefit-type comparison of the alternative approaches is not always easy. While open-source software is free to download and adapt (and usually to redistribute) that doesn't mean there is zero cost of ownership. Far from it. What is saved on up-front costs and ongoing licensing may be spent on support and integration and special features to make the software enterprise-ready.
"Those economics need to be thought about, otherwise you create this false expectation," says Bob Laurence, CEO of Crunchy Data. "But what's developing is a market in the true sense of the word. You've got little groups of users who have been wildly successful using open source, now they want to bring it into the enterprise and they ask ‘how do we know there's expertise out there to help us? Who do we turn to?' Those are the sort of questions you don't have to ask if you use Oracle."
Laurence insists that even with support costs factored in, most enterprise open-source projects cost much less than proprietary alternatives, but says that companies should not skimp on the support. For mission-critical use cases, he goes on, "you need someone who is there for you with the right expertise and the right responsiveness. If you're a major bank you can't just rely on the [community forum] for help."
Such issues come to a head with big data, which is overwhelmingly based on open-source software and where the required expertise is thin on the ground, and they also apply to data management and security features. For other modifications, though, companies may prefer to deploy their own developers.
Lane is unsure whether or not developers for open-source projects command higher salaries, but he does believe that many engineers, especially younger ones, prefer working on such projects simply because their work is acknowledged by their peers rather than being hidden away. If staff retention is an issue for a company, this may be something to consider.
Making the most of your developers
It is also true that open-source projects are generally more dynamic, with improvements and patches coming in from all sorts of directions. Companies can draw on outside help without necessarily paying for it, and updates are often more frequent.
"If you are a software developer and you want to get involved with the Postgres project, the reason you are doing that is you need a database and what's out there doesn't do quite what you want, so you're either faced with writing something from scratch or trying to extend some existing project. That's pretty much where we get all of our development from," says Lane, who adds that increasingly companies are seeing the benefit of paying developers for time spent contributing to open-source projects.
"Companies are seeing that this is a reasonable way to approach things. They can leverage their development efforts by co-operating with other people who are contributing to a given project. It's that positive cycle of contributing expertise and improving the software which results in a great product."
Can you rely on loss-making vendors?
But what about the wisdom of relying on companies that are not profitable? In contrast to Oracle and Microsoft, most open-source firms, even long-standing ones, barely break even - if that. One of the fastest growing software companies in history, Hortonworks, is still loss-making. There is less danger of lock-in with open-source but you can still be left high and dry if funding or develper interest drops off taking with it the regular patches and updates on which business software depends.
In this context "'turning a profit' is a squishier thing than it might appear," says Lane. "My former employer Salesforce isn't showing a profit, though they are taking in money by the truckload."
Lane goes on: "I remember asking a VP about this when I was interviewing there, and he explained that they could show a profit anytime they wanted, just by letting off the gas a bit on hiring and expanding, but they preferred to grow as fast as they could instead. That's probably the same with Hortonworks."
He continues: "It's probably smarter to watch their revenue growth rate if you're wondering whether they have a sustainable business."