Of all of the Fortune 500 enterprise architects out there, The Hartford’s James McGovern is hands down my favorite. Not, as might be suspected, because he’s been kind enough to mention RedMonk on a regular basis.[1] James and I appreciate that, of course, as we do whenever someone takes the time to let us know how we’re perceived and performing. But of greater importance is his distinctly unconventional willingness to consider alternative solutions (here are his thoughts on avoiding groupthink) – something other than the “nobody-got-fired-for-buying-x” choice.
The iconoclastic behavior on his part manifests itself in all kinds of creative and interesting ways, from conversations we’ve had with him on topics such as COA, to his comments on the industry analysis business, to his thoughtful analysis of the role of open source within the enterprise portal application space. He’s aggressive in his commentary, and may make many vendors and industry analysts uncomfortable with his candor, but if you’re not reading him and you’re interested in the practice of IT within the enterprise, you’re doing yourself a disservice.
While he’s occasionally quite declarative, McGovern also takes the time to ask some very interesting questions – questions that are important ones, IMO. Matt Asay and the gang from Alfresco, for example, might want to take a stab at some of the queries regarding ECM asked here; I’ll hopefully chime in on a couple of those myself eventually.
But for now, in the spirit of open source analysis (and b/c I’m a sucker for Q&A’s), I wanted to address some questions McGovern directed our way a little while back. The questions are included below, with my responses. Both McGovern and I would certainly appreciate any other thoughts you all might have on either the questions or my responses:
- How can human resource departments find talented people from the open source community so I can support open source internally?
Great question. Frankly, I think it would be tough on HR people for a couple of reasons. First, because it depends on what kind of open source people you’re looking for. Say, for the sake of argument, you’re looking for some talented devs in the database space – there are at least half a dozen open source database projects of significance, and maybe a dozen or two that are relevant in some way to an enterprise. Further, the projects are quite different; berkeley db is not db4objects is not Derby is not MySQL is not Postgres is not SQLite and so on. Where would HR start? Second, the types of place you’d have to look may seem a bit intimidating to many non-technical HR staffs – IRC, for example, probably doesn’t mean much to them.
But fortunately, all is not lost. The shortest distance between your problem and the answer is someone who already knows it. There are resources available – such as RedMonk – that could probably answer or get answered many questions about talented people and/or firms in the open source space. If asking for help is not an option, the next step is monitoring some of the same channels we do: blogs, listservs, forums, and wikis. They’ll go a long way towards helping identifying knowledgable resources. Shortcuts to that process would be looking to those that have been granted commiter or PMC status on projects in an area of interest to the enterprise. Less ideally, you could track awards given out at hackfests, open source conferences, etc.
- Which corporations whose business is not technology are currently contributing to open source projects?
Will have to ponder this one further, as I don’t have any catchy examples top of mind, but I’d look at open source – or in the case of the JCP, other participatory bodies – that have substantial industry participation. Eclipse, the JCP, Liberty, and the WS-I all come to mind here. The JPMorgan/AMQ idea is another potential project worth looking at. But I suspect the question is more aimed at soliciting ammunition to prove that non-vendor enterprises are implicitly fueling development by indirectly sponsoring development – I’ll have to think on that one a bit. James, you might want to follow up with Optaros’ Stephe Walli for examples there, as he’s worked on that quite a bit.
- Which open source projects are pervasively implemented in my particular industry vertical?
It depends on how you define pervasive, but with maybe one exception, every insurance company I’ve worked at or spoken with has deployed both Apache and Linux in some capacity – though a quick Netcraft query of firms like State Farm and Country Financial indicates that both are IIS shops. From there, it gets much more difficult to assess; I’m aware of substantial MySQL and JBoss roles in insurance shops, but would not in any way characterize them as pervasive. On the tools side, Eclipse was the IDE of choice at the insurance oriented Java shops I’m familiar with.
- Where would I expect to see non-commercial open source projects beat proprietary vendor products?
Answering this requires a clarification on what the metrics are for winning. Beat proprietary on price? Or performance? Features? Ease of use? Marketshare? Pricing is one area that open source can – but is not always – beat proprietary software. This is not necessarily because the software is “free,” per se, but rather because of a variety of other factors: the open source alternative might be more entry level (see the database market), the cost of the software development might be ammortized over multiple organizations or individuals (see the Linux kernel), the cost of the software might more accurately reflect usage (i.e. paying only for instances that require support), etc. But proprietary can also be quite competitive on price.
On a performance basis, some of the aforementioned projects have traditionally performed well: Apache’s proven itself sufficiently (though IIS has made great strides of late), I think, and Google certainly seems relatively satisified with its investments in Linux, MySQL and other open source projects.
Ease of use could be argued in either direction, I should think. On the desktop most would cede the point that Windows or OS X are simpler to use than the open source alternatives (yes, I’m aware that OS X is based on an open source kernel), if only because of the vastly superior available hardware support. In infrastructure software, on the other hand, many of the open source projects are simpler and easier to work with – if only because they are often less functional than their proprietary cousins. The database world is perhaps the best example of this; I talk to a great many developers that use MySQL not because it’s open source, but b/c it’s easily obtainable and relatively easy to work with (if I can figure it out, anyone can).
But if the question is a general one, as in: where is open source the best value versus proprietary alternatives, I think my answer would be in the commodity infrastructure space (OS, RDBMS, app/web server/etc), simply b/c those are volume opportunities that allow open source to shine. Are there exceptions? Certainly, on both sides. But one need look no further than the success of the LAMP stack to appreciate just how successful open source is in the infrastructure space.
- What are some “best practices” for corporations who would like to contribute to the open source community?
I could probably discuss this question endlessly, but the one that I always come back to is deceptively simple: don’t just take, give. The percentage of commercial entities that actually give back to the projects they utilize – whether it’s in the form of code contributions, or other rewards – is low, when compared to the volume of firms that rely on it. But this is shortsighted on their part, for reasons that are not particularly complicated.
Let’s take code as an example. When working with an open source software, it’s not terribly uncommon for members of more experienced IT staffs to gain at least a working knowledge of the source involved, and occasionally identify potential fixes or even enhancements. If you don’t buy that, ask yourself why the assembled CIOs here care so much about access to the source. Once the fix or feature has been identified, however, there are three basic choices:
- Make no modifications to the product, for fear of jeopardizing a support contract and/or making future upgrades more problematic
- Apply the modification to your code only (even the GPL permits this as long as you’re not distributing the work), thus commiting yourself to reapplying the patch in every future upgrade (and hoping it works), while jeopardizing your support contract – and depriving other users of your fix in the process
- Contribute your patch upstream to the project; if said patch is accepted, which unless your staff has commiter status is not guaranteed, you won’t have to worry about either support or future upgrades – and simultaneously you’re helping to ensure the ongoing viability of the project in question
I don’t know about you, but I know which one I’d pick.
- How can corporations participate in creating open-source industry analysis?
Several different ways:
- Ask Questions:
Those of you who’ve been reading for a while will be unsurprised to learn that I believe that most analysis starts with questions. What do enterprises need to know? What are the questions they’re struggling with? Analysis doesn’t occur in a vacuum, so the first step is having someone ask the right questions. Not all will be answered, but some might – and just the exercise of asking has its value. - Answer Questions:
One of our core beliefs at RedMonk is that the practitioners – the developers, the architects, the engineers, the designers – are the real experts. While I try to install and play with as many of the technologies I cover as possible, I simply don’t have the time to devote to the products that someone running it in production will – therefore it’s those opinions I rely on to augment my own experiences. In other words, many of the folks looking to benefit from open source analysis should themselves be participating in it, because they have answers others are looking for. So if you run across people asking questions that you know the answers to – whether that’s in blogs, forums, wikis or otherwise – try and take the time to answer some of them. You can see McGovern doing just this over on Friend of RedMonk Brenda Michelson’s blog here. - Share Your Experiences Publically:
At some level, industry analysis is the simply the aggregation and distillation of customer experiences. When a customer is weighing potential solutions, the intelligence they most desire is simple: what are the experiences of lots of previous customers? While the traditional, proprietary industry analysis model depends on private solicitations (and subsequent regurgitation) of this information, that makes the analysis less than open. The best solution to this is to encourage firms to explicitly publish such feedback so that the conclusions that are drawn from it are transparently derived. - Be Willing to Pay for Value Received:
Had to drop this one in there – I do like getting paid 😉
- Ask Questions:
Disclaimers: Of the above firms or projects, db4objects, Eclipse, Liberty, Microsoft (Windows), Sleepycat (Bekeley DB), and Sun (JCP) are RedMonk clients, while Alfresco, Apple (OS X), Google, and MySQL are not.
[1] Even going so far as to call us out explicitly during a panel he sat in on at OSBC.
