i5/OS and IBM i Support: How Long Does It Last?
January 17, 2011 Timothy Prickett Morgan
Now that IBM has given i5/OS V5R4 a stay of execution and extended its marketing life out to May 27 of this year, the question that everyone now wants to have answered is how long will that venerable operating system for prior Power-based systems have support from Big Blue? It is debatable how much history is any guide, since the IBM i platform is not at a particularly strong point in its upgrade curve for a lot of different reasons.
As you can see from the official OS/400, i5/OS, and IBM i support matrix page, IBM has not yet talked about when it will pull the plug on PTF patching and technical support, offered through its Software Maintenance (SWMA) service, on i5/OS V5R4. Nor has it said when Program Support Extension services, which offer support beyond this date for an incremental fee over and above the normal SWMA (which you still have to pay) will be available on i5/OS V5R4 and when that service will sunset.
As is the norm for fairly new versions and releases, IBM has not told customers when it will stop selling the more recent IBM i 6.1 or 7.1 releases, and it is a long way from getting close to talking about end of support or end of program support extension. All IBM ever says about these things is that it will let you know when support will end on a version and release at least 12 months prior to the cutoff point.
Back in October, Joe Hertvik, our resident Admin Alert columnist in the Four Hundred Guru tech tips newsletter, took a stab at when IBM might stop offering support for i5/OS V5R4, and he reasonably guesstimated that V5R4 would see end of normal support in April 2012 and then program support extension until April 2015. (You can see his reasoning for these support estimates here.) With V5R4 now being sold through May 27, 2011, instead of being withdrawn on January 7, 2011, you can add 140 days to Joe’s estimates, which puts normal support closing down in late August 2012 and extended support turning off in late August 2015. That’s assuming that IBM was going to move these in synch, which is not necessarily a valid assumption. If these numbers hold, then V5R4 would have been under normal and extended support for a length of time that far exceeds any other release in the history of the OS/400 and i platform.
That’s one reason why I don’t think V5R4 will be given such a long life, which is about six and a half years for support from general availability to end of normal support, plus three more years on top of that for extended support.
Let’s take a look at the OS/400, i5/OS, and IBM i support matrix and do a little math on it in this patent-pending TPM table. In this table, I took the basic information that IBM gave in its Web pages and added in the extended support data (what I could find, as you can see, I have some holes in my data), and then calculated the number of days between announcement and general availability, and then from general availability to other key milestones in each release’s support lifecycle.
The first thing you will notice in this table is that the time from announcement to the general availability of an operating system release is all over the map and has been since the AS/400 was launched in 1988. This just goes to show you that market forces and the difficulty of delivering an OS version and release determine when new software gets released. Sometimes IBM is ready, and sometimes it is not but it needs to get something out the door for marketing and sales reasons.
The next thing you will notice is that the economic lifetime of an OS release–from general availability to the end of marketing–has bopped all over the place, but general speaking is just under three years. V5R4, after two marketing life extensions, is now slated to have a lifetime of more than five years (1,928 days, to be precise). The V4 releases came rapid fire, with lots of new technology, and were short-lived by comparison with the V3 and V5 releases. V6, also an even-numbered version, could have the shortest life of all considering the overlap with V5R4 and IBM i 7.1. If you have to jump to i 6.1 and suffer through program conversion or pay for ISV software upgrades, you might as well go straight to i 7.1. And if you can’t afford the ISV software upgrades to move ahead, you’re going to stick to V5R4 and older iron and not move to i 6.1 just for the fun of program conversion.
If history is any guide, then IBM will announce end of support for V5R4 soon, and it will likely run out in early 2012 as Joe had been expecting. Program support extensions could be offered for as little as two years beyond that to as much as four years, depending on how much effort this takes IBM to do, how much money IBM can extract from midrange shops stuck on V5R4, and what Big Blue does to make the jump from V5R4 to i 7.1 easier.
I have said this many times already, but it bears saying again. In this radically virtualized server world we now live in, the answer is to not force companies to have to shift off V5R4 at all to move to modern Power7 iron. The answer is to run a virtualized version of V5R4, including unconverted RPG and COBOL programs and databases, in a logical partition atop the PowerVM hypervisor.
As long as the iron is cheap enough and the PowerVM and i 7.1 licenses are priced reasonably, customers will be happy enough to run in emulation mode. This worked for the System/36 and System/38 environments and their applications back in 1988, after all. Sure, there was some grousing. Sure ISVs didn’t make as much money on software upgrades as they liked. But at least this way, customers moved ahead to the AS/400 and got on the path toward modernizing their applications, and that was far better than having them all dig their heels in.