Making Billions with Open Source, Revisited

Every quarter or so, there’s an old can of worms that we with a foot or two in the open source business world open up and dump on someone’s head. Dave Rosenberg appears to be getting out the can opener:

Customers assume that open source is free and that commercial open source is cheaper, but most companies aren’t prepared to deal with the implications of having a lower cost product. Even when you can clearly demonstrate value, you run into a scale issue sooner or later.

So, the answer appears to be that you have to provide more value for the dollar, but how do you do in a way that makes a highly scalable, highly monetizable business?

This is the question of how someone gets rich off an open source business. Typically, the phrasing goes, “how does a business make money with open source,” but my rephrasing there gets at the first problem. First off, let’s pull the zombified dead-horse out of the ground and examine it:

Paying for “Free”

The issue with open source software business models is that your customers can get your software for free, without ever paying you or even (we hope) registering with you, giving you lead information to call them up and pester them. What do you sell them then?

Support & Services

Support and services is the first rung of “products” you can sell without muddling with the definition of open source. “Support” being answering the phone, email, or even being on-site to trouble shoot a customer’s problem using the software. Typically, people wrap metallurgical bands around this – Gold, Silver, Platinum (whither “Lead”?) – that scale up the response time (2 hours to days), availability (business hours, weekends, 24/7), named contact (“I talk to Jim about this, not some random call-center head-set”), and other features that don’t directly connect to the software itself. RedHat pulled a clever trick many years ago here: instead of selling you the source code, they sold the binaries, the compiled code. It was certified and approved by them and the first step in getting all those wonderful support options.

“Services” are butts-in-seats things like “make this open source software work with my SAP install.” That is, integrating and customizing the software to do whatever business function the customer needs. You could go and download JBoss App Server and the Spring Framework all day long for free, but you’ll need to actually code something if you want to start making money with them.

There’s tons of money in “services” like this, but they’re not typically what open source businesses go after. Margins tend to be bad (compare the cost of a living, breathing human to copying software), there’s tough competition, and there’s an overall distaste among ISVs to actually customizing their software per customer.

Dragons Be Here!

After this, things get a little tricky, and it’s when you cross The Tarus Line. Going past The Tarus Line means that Tarus Balog, noted, and more importantly, successful open source business person will start telling you you’re not doing open source.

Patches & Updates

First, open source vendors can be paid to supply patches and updates to software. All software is full of bugs, security problems, and other issues that need to be fixed. Shockingly, even open source applies here. Less glibly, new features are added to software all the time. In the closed source world, we’d call these new versions of the software. The open source world is a bit different than bundling up mega-released: there tends to be more incremental releases, or updates. Either way, customers want both bug fixes and new features.

Selling quick, easy, and even early access to those updates is another monetization point for open source businesses.

The philosophic “issue” here is that the underlying code for those updates should itself be open source as well, so why can’t the customer just get ahold of it, re-compile things, and deploy them? Certainly, there is value in paying someone to just do that for you – the vendor – but the value in that shouldn’t be high enough to get to the “highly scalable, highly monetizable business” goal that Dave’s can-opener is ripping away at. How much revenue can any given company depend on for simply providing the service of updating open source software?

Now, if you’re a clever company you start to wrap things together here: you’ll only get that support if you use those certified updates. If you went and re-updated yourself, we – the open source vendor – are going to hassle you with a sort of metaphoric “warranty void if removed” argument against helping you.

Again, the issue is not if this is good, bad, open, or closed source. Rather, the question is how much revenue any given company can get from such a service.

SaaS and Cloud

Another option is to charge for actually running the software for your customers. You put the software up in the cloud and charge for the service you’re providing instead of the software. Google is a sort of square peg in a round hole here – they don’t open source all of their code – and you’d think people like Sun would be a natural fit for, say, running identity services based on their open source software in the as a service (or, yes, “in the cloud”).

WordPress is, perhaps, a more perfect example. You can download WordPress for yourself and run it (as we do at RedMonk) or you can pay to simply have WordPress.com run it for you. The interesting part here is that the WordPress instance you get at WordPress.com – that you pay for – has less features than the one you’d run for yourself. The hassle of running your own blog (and elastically scaling up when you get lots of traffic) is worth not only paying for, but axing features and flexibility.

The problem (or pointer to what needs to change?) with the WordPress example is that it’s not purely “enterprise” or “business” software: it’s consumer as well.

In theory, there’s good revenue here. Closed source SaaS’s like SalesForce are stand-up businesses. There aren’t a whole lot of open source SaaSes out there.

On a related note, we here at RedMonk have been trying to develop a theory of how you can monetize the community around open source stacks via telemetry and collaborative IT management (or “social business software” to generalize it) for sometime now. Ask us for more info (check me out: monetizing free content!)

There’s another trick, of course, as well here: often, to run software at large scale as a SaaS you need some additional code to help out. Be that code to manage it, balance out databases and web servers, whatever. The code that Google, for example, would be insane to open source. When looking at an open source SaaS, you have to make sure to ask them if all of the code is open.

Lump It: Sell Closed Source

Finally, getting to the buried lead, an increasingly popular open source business model is to sell closed source. This closed source software typically comes in two forms: additional software used to manage the open source software or “plugins” and “extensions.”

In both cases, typically, the closed software is being sold to someone other than developers. And this, I believe, is key: developers will not tolerate paying for software (except Microsoft developers), pretty much of any kind. The people who run the software day-to-day – admins and operations people – will pay for software.

My pet software category, “The Little 4” (tho, now there’s just 3, really) fit into both of these categories.

Zenoss, Groundwork, and Hyperic are all open source businesses of varying types. In each case, though, when you pay those vendors money, you get closed source software. In the case of Zenoss, you get additional functionality in the form of “ZenPacks” that allow you to monitor and manage more things in your data center (like VMWare). In the case of Groundwork, you get the pulling together and polishing of all of the open source project they amalgamate, and also additional functionality. Hyperic is the same.

Additionally, Hyperic is an interesting example because they’ll license white-labeled versions of their management software to other open source vendors. For example, if you “buy” Mule from MuleSource, you’ll also get a version of Hyperic’s HQ tuned to monitor and manage your Mule install.

Compare this to, for example, what you get from another outfit in the business of IT management, the OpenNMS Group (home of The Tarus Line): all of their software is open source. All you get from them are services and support.

More People Selling Closed Source

Other examples of selling closed source exist in outside of the IT Management category and I’m increasingly seeing more of it. Personally, I see no problem with it. If customers will pay for closed source software, bully for them and the vendors. Software is valuable, after all. Getting email and instant messaging working for your 10,000 employees has value, as does munching over 1,000’s of money making transactions an hour.

Far from running around waving a flag of “open source is dead!” I’m suggesting that vendors should start feeling (less bad) about selling closed source software. Certainly, there’s an ethics to trying to hide that fact (something that The Little 4 people above do not tend to fall into the trap of doing) and benefit from the marketing glow of being “open source.” No one wants hucksterism here, folks.

But, making a Big Buck off open source is hard. Closed source tends to be easier.

But what about MySQL?!

Indeed, Sun bought MySQL for some Big Bucks. RedHat bought JBoss, and so on. In general, I don’t consider “getting bought by another company” a business-business model. It may be an investment and personal wealth building model, but a 20, 50, 100 year business model it ain’t. Clearly, an open source startup with the goal of cashing out after several years is a viable business model for the people who own a lot of equity in that company. But that’s not the kind of business we’re talking about here. There is no “cashing out” for Sun, IBM, Microsoft, or Oracle: just the quarter-by-quarter game of growing a few ticks or taking a beating from Wall Street (love those Street guys! Kisses!).

It should also be noted that the owners of previously independent open source companies tend to use all of the money extraction techniques above, esp., the grand-daddy example here, IBM. IBM sells plenty of closed source software, hardware, services, and just about everything but mercenaries. And yet they do plenty of work with open source as well.

What Should I do?

As a customer, you should get ready to start paying for more open source software. As a vendor, you should start thinking about selling closed source software in addition to your open source offerings. These rules of thumbs don’t apply to all categories: when you get to more middle-ware, developer-centric software stacks, it’s an extremely tough sell to get people to pay for software. Then again, someone managed to invent the bottled water market which looked like insanity – charging for “free” water.

Open source will never go away as a method of production or a business model. In fact, you’d be insane not to use it as a core part of your software business model now-a-days; don’t confuse the above with me saying otherwise. Indeed, as the OpenNMS Group and other “small” outfits have shown, you can build a fine business of “pure open source.” It’s those “highly scalable, highly monetizable business” that are a tougher nut to crack and a little help from closed source can look like a big mallet.

Update: see this addendum on models that involve using, or “embedding” open source.

Disclosure: IBM, Hyperic, Groundwork, Zenoss, Microsoft, Sun, and RedHat are clients.