Red Hat yesterday unveiled its answer to the vexed question of what it should do about its consumer line - dump it. This is not quite how CEO Matthew Szulik put it to The Register over lunch yesterday, nor indeed did he say flat out 'oh, by the way, we're shooting the Red Hat Linux distribution when North Carolina wakes up, bye now,' but that is indeed what the company did.
Red Hat maintenance and support for Red Hat linux 7.1, 7.2, 7.3 and 8.0 will end as of 31st December 2003, and for Red Hat 9 as of 30th April 2004. "Red Hat does not plan to release another product in the Red Hat Linux line," says the 'you're fired' email sent out to The Register's Red Hat Network account yesterday. Red Hat Linux channels will remain open for six months after the product's end of life date, but no new errata will be posted after EOL, and no Red Hat Network support accounts could be bought or renewed as of yesterday. So it's very dead, very soon, and Red Hat is now about Red Hat Enterprise and... Fedora.
The move is a sensible one from the company's point of view because it provides a clear answer to the 'what do we do about the bit that doesn't work' question. It could perhaps have been done a little less brutally, and people who bought RH8 or RH9 both surely have good reason to be sore about how swiftly the support is vapourising (RH8 in particular - a year ago this looked like a product line with a future and a roadmap), but there really was no easy way to do it. A more honest way, probably...
What we have now are two lines, which you can term commercial (Enterprise) and developer (Fedora). The strategy for the Enterprise line has been pretty clear for some time now; it's the paid-for line that started with high end products but which has expanded downwards to smaller servers and workstation. It makes money, and 'buy workstation' is what Szulik tells us the people who'd been previously buying RHL should do in the future (although as we suggested earlier on, he didn't make it entirely clear how close that future actually was).
To some extent, selling workstation through retail channels provides some justification for claims that Red Hat hasn't got out of consumer or retail, but as it's just killed off a product line it surely must have got out of something. Or not. Red Hat's problem with RHL was that it wasn't associated with a clearly-defined 'something,' whereas Enterprise was and is. Hold onto the fact that Red Hat will be pitching a business focused product into the space vacated though, that this won't take up all of the spaced, and that the company is not proposing to have a product that takes Windows head-on in the general client OS market.
There's a possible gap there, one that Red Hat wasn't able to figure out how to address right now, and we'll likely be able to gauge how big it is as we see how hard and how much rival distros try to fill it. SuSE, for example, sees Red Hat's perceived lack of commitment to the enthusiasts as a major vulnerability to be exploited, although it's currently not about to launch a client-based consumer assault on Microsoft either.
Fedora is Red Hat's bid to cover this base, but frankly it would look a hell of a lot better and more convincing if RHL could somehow just magically disappear, and never have existed in the first place. By pulling out of RHL Red Hat has given Fedora a major credibility hill to climb. Which is a pity, because you can see how it might work in a parallel universe.
Fedora is the free, development strand of the Red Hat offering, "sponsored" by Red Hat but intended to be out there in the community with support almost entirely from the community. It will have a release cycle of two to three a year, and Szulik puts it more bracingly than that still. "Anyone can put a package in so long as they maintain it," and fixes will "just roll forward to the next version," so it'll ship when it's done and keep shipping. It is therefore being pitched as a fast-moving, cutting edge line that's in continued development, and in that sense could be seen as a possible techie heaven.
Case for the prosecution though. The support headache is being largely offloaded from Red Hat, and although Szulik says there are plenty Red Hat people involved, it seems fair to us to define their involvement as that of enthusiasts - it is not, in most cases, their day job. Developments from Fedora will go into Enterprise, so to some extent Red Hat will be getting development for free, and is throwing what's left of the old model back into the community while it concentrates on its ROI. The issues here are more complex than that, of course, but we did say this is the case for the prosecution, right? People are going to think and say these things, and it will be something Red Hat will have to deal with.
Fedora's development and release cycles and its support mechanisms clearly aren't going to be any kind of answer for people who want to run a business on Linux, whereas Enterprise's more sedate cycles, support model and product lifespan are. This is certainly a convenient piece of ringfencing for Red Hat, but you can't really count it as entirely deliberate. Businesses want long, defined lifecycles, clear roadmaps, support, while enthusiasts want fast development, so the convenience factor is built into the two logical answers Red Hat has come up with.
But it could have done it, and put it, better. Might we suggest telling customers, 'Sorry, we made a mistake, this is what we're doing now, this is how we propose to make it up to you' would have been a good place to start? ®