Red Hat's new support policy for operating systems represents a long-delayed tidying up, and is only - as some of you have suggested - part of the whole picture, says company VP, product strategy, Erik Troan. At the moment, the company has a three- to five-year policy (which is being 'clarified' to a minimum of five this week) for its flagship server product, Advanced Server, and a one year policy for its 'consumer' products.
Which at the moment is everything else, which is really where the problem lies. Red Hat currently has one big, fairly expensive server product aimed at business, and then the 'normal' stuff. There is clearly a hole for people who want to equip business desktops or deploy smaller servers, and want to do it with Red Hat. Troan confirms that the company is moving to fill it.
Red Hat Advanced Workstation will be out later this year, and the company also proposes lower cost versions of non-consumer server products "that fit in below Advanced Server", which should give the company a clearer and more viable product range, with consumer being the traditional open source stuff you can get for free, and that updates eye-wateringly fast, while non-consumer has upgrade cycles and support periods that are in line with businesses expectations of being able to deploy something and have it supported without major upgrades for three to five years.
"Three to five," incidentally, was the period Red Hat talked about internally prior to specifying the support period for Advanced Server, which is where the published "three" minimum came from. Five is more sensible, at least from a comfort point of view (because in the real world quite a lot of people will upgrade before this), but in many cases the actual period of guaranteed support will be much longer. Key accounts, says Troan, will want and get in excess of five, while defence will typically want way, way beyond that.
The acid test will be pricing on the forthcoming non-consumer products. Troan suggests it won't be a huge amount, but given that Windows comes free as far as most users are concerned (with an actual cost of maybe up to $50), there isn't a huge amount of leeway on the desktop. Red Hat could compete with Windows prices if it were selling multiple licences to business, but couldn't undercut its own consumer product at retail. Given the mare's nest that it Microsoft's pricing at retail, we do not propose to go there - hardly anybody buys Windows retail anyway.
On the other hand, what if the nature of retail Linux product were to change? Would there be a wider market for a non-geek version of Linux, aimed at people who just wanted to get stuff done? That might actually turn out to be a more appropriate packaging to sell at retail, with standard/consumer versions maybe switching to electronic sales only. But as with practically all of desktop Linux today, this is uncharted territory, and it's stuff we speculated all by ourselves, without Erik saying anything to suggest it.
As regards the support period of 12 months for consumer, Red Hat does have a point, and it's a problem all of the Linux distributors will face if they're not going to slow down. Red Hat puts out two versions of the consumer line a year, and it's clearly difficult and expensive to support these products for five, or even three years, particularly as quite a lot of users don't actually pay Red Hat anything for the software. Troan argues that the Red Hat Network, which provides automatic updates, is a factor here. The more you have to support, the longer it takes to get security updates to users. In addition, upgrading versions in Linux is in general a lot less onerous than doing it in Windows.
Red Hat's data says 90 per cent of users of the Red Hat Network are running software less than a year old, and Troan says use by older versions flattens right off from the launch of a new one. So he argues most users do update, and anecdotal evidence tends to support this; most Linux users grab the latest just as soon as it's out, don't they? They're like that.
Earlier End of Lifing does however present problems for you if you happen not to like some things about the more recent products. You might not, for example, like Bluecurve or the MP3 policy of Red Hat 8.0, or have more serious objections regarding choice of file systems, as some people do. Specifically, as regards 8.0, there seems to us to be a slight strategic mismatch.
With 8.0, Red Hat introduced features intended to make the product more generally accessible to the unconverted, and this would seem to us to fit more readily with a longer-supported, less frequently upgrade product line. That is, not the consumer line, the other line that is yet to arrive. But it's work in progress, we accept that. ®