In this article, Sun Chief Technology Evangelist Simon Phipps states that the subscription model is a “necessary trend for Open Source Deployers.” I disagree, at least where the JDS is concerned.
Phipp’s key analogy is that of a newspaper:
The model Sun is developing for giving customers the benefits they need using open source is illustrated well by Sun’s Java Desktop System (JDS). JDS comprises many software elements drawn from a wide range of open source communities. To understand what’s happening, let’s consider the newspaper industry. Newspapers haven’t been killed off by the Internet (at least, not yet!). The reason for this is that when we buy a newspaper, we’re not buying the news. These days the news is free (gratis) – we can go online and read news feeds from organizations like Reuters or the Associated Press, or original reporting from an organization like the BBC. When I buy a newspaper like ‘USA Today’ or ‘The Independent’, I am actually buying an editorial style. The editor-in-chief for the newspaper sets the outlook, and then the editors and other staff select the news stories, phrase the reports, position them in the publication and perform the lay-out in support of that editorial outlook. If I go online to get the news, I have to do the work of selecting and filtering the news, and I may not always be aware of the biases of the source I am using. To get an aggregation of the news I want delivered in a style that helps me and with biases I understand, I subscribe to a newspaper. JDS is just like this. Almost all the elements that comprise it – the Mozilla browser, the Evolution mail and calendar client, the StarOffice document productivity suite, the underlying GNU/Linux operating system they depend on, the Gnome desktop environment they use and much more – come from open source communities. You could go get all those parts yourself – they are all available gratis. But then you’d have to integrate them yourself, support them yourself, and accept joint liability for their use of ideas yourself. Instead, Sun acts like the editor-in-chief of the JDS ‘publication’. Staff select the software components to include and exclude, work to integrate them, contribute to each of the open source communities to improve their compatibility and completeness. Sun packages and delivers the final publication, offers support and updates, fixes security exposures, offers indemnity and generally joins the communities so you don’t have to. You don’t buy the software from Sun – instead you subscribe to the editorial outlook. Sun’s editorial view is to deliver high function, ease of use, data format and networking compatibility, low migration cost, re-use of existing hardware, escape from Windows viruses and security risks and minimal retraining. If that’s an editorial outlook that fits your corporate needs, you’d do well to subscribe.
The problem with this notion is that the newspaper itself — with its editorial outlook intact — exists online for free as well; online news seekers aren’t limited to raw feeds with unknown biases. If I want, I can read the Washington Post in its entirety — or the Washington Times, if I prefer my news with a Moonie bent. In fact, most papers are available free online, though many require registration. What we buy with a newspaper is portability, which isn’t that germane to the subject Phipps really wants to discuss.
The notion of JDS as a known aggregator is sort of buyable, but there are distinct drawbacks to working with someone else’s idea of how these pieces should work together. Unhooking them to change the configuration will require nontrivial work, so the time you spend using someone else’s aggregate may be false economy. PAYING Sun in the long run for this desktop may be a way to get into FOSS quickly, but ultimately you’ll want more control, and you’ll go to a homegrown solution. And because you can’t really escape having SOME IT at your enterprise grows, eventually it’ll make more sense to develop your own build in house and stop paying Sun every month.
This is analagous to the hosting situation of a new firm, for example. Right now, say newcompany.com lives at Hostcentric on a server they provisioned and built to their spec. It has most of what we wqnt on it, but in a configuration that isn’t optimum for adaptation or tool migration — if, say, we wanted to change our mail system to authenticate to LDAP, for example. When we start wanting to do fancy things, we’ll need to build our own server so we can (a) pick the tools we want to use instead of the ones Hostcentric likes (b) control the way these tools interoperate.
Sun has a big challenge ahead: it needs to define why it still exists. High bus usage applications where their tightly coupled hardware/software solution shines are a tiny portion of the market they formerly dominated; there’s no longer a need for most applications to go to such high-end, proprietary hardware. For example, last year I led a team building a music marketing site that now supports 150,000 registered users and 3MM hits a day. In runs on two Intel boxes (one runs Apache; the other runs Postgres). They’re beefy boxes with RAID and lots of RAM, but they cost a fraction of what it used to take to support that kind of application, and that means most of Sun’s market has gone the way of all flesh.
What does Sun have? They have Java, but monetizing that would be difficult at best. They have Solaris, which may be the last man standing in the proprietary Unix world, but Linux is closing fast, and being the last survivor of a dying breed still makes you a dying breed. The JDS is an interesting idea, but isn’t that compelling, I don’t think, if you look more carefully at the way it purports to work. We get a Linux machine with free tools installed, and we pay Sun to manage it for us? Um, why? How much? First, Linux boxen require dramatically less “management” once running than do Windows machines. Second, why would we want to entrust any outside entity with the stability of our machines? In a small organization without any IT support, I can see a degree of appeal here (again, like the Hostcentric situation), but any sort of ongoing and growing enterprise will likely be better served by assembling a “drive image” internally and distributing it in much the same way Windows is managed now. Sun is adding a financial “load” here without clear value-add, at least from this geek’s point of view, and adds an external point of failure besides.
So how do you make money in open source? There are lots of ways, but chief among them are probably service model businesses (somebody’s always got to know how to make things work together no matter who made the software) and open-source-based projects like Apple’s OS X (built on FreeBSD), Tivo (which runs Linux), and a thousand other tools built with free or open source cores. Trouble is, I’m not sure how Sun can make that jump. JDS is an attempt, I suppose, to product-ize a service (that is, managing your Linux desktops). The problem with this is that I have a hard time imagining they’ll be able to charge enough, or be able to get enough customers, to make it play for them. They need a LOT of revenue to avoid being a niche player, and I don’t think this is the source. I’m not counting them out yet, but I wouldn’t buy the stock, and I sure wouldn’t trust my company’s desktops to a firm whose business model seems on the way out the door.