Comparing next solutions for 3000s
HP releases critical patches to prevent corruption

Clock up patched support's value

HP still has about 48 hours to go before it might reply to OpenMPE's request to open up the HP 3000's source code. Only one value stands out as the chief benefit for such a license of MPE/iX to a third party: patches, and the ability of a non-HP entity to create or modify such modules of the operating system.

As of this morning, patch delivery and creation looks like the most obvious difference in service levels between HP's 3000 support and that of third parties. Even then, patches that are already released — beyond beta testing — can still be downloaded and used today, even if HP never begins its source code transfer for MPE/iX.

OpenMPE wants HP to announce something by Nov. 1 about starting the source code handover. The transfer should take about a year, by everybody's estimates. The OpenMPE advocates say that patch creation will be important to the 3000 customer who operates the system beyond December of 2008, when HP ends its patching operations. That means getting ready to patch should begin by the end of 2007.

But this weekend illustrates one of the few instances where a patch is necessary to run an HP 3000 safely. On Saturday evening around much of the world the clocks roll back, away from Daylight Saving Time. On a new weekend, for most countries. This is what passes for a critical patch in the days when many HP 3000s are locked down, frozen with few changes allowed.

Many third party support companies refer to patching any 3000 as a last-resort strategy. This is no slam against HP's engineering capability, but the belt-plus-suspenders credo which built the 3000 into the industry's most reliable business server. Any workaround, support companies say, brings a lot less chance for disruption than a patch.

HP was careful to note in its announcement last month that not even security-related patches will be developed inside HP labs from 2009 onward for the 3000. Those patches are rare, too. Many of the Denial of Service kinds of exploits won't cripple an HP 3000 like they might an HP Unix system. HP built things like Domain Name Services on a different OS architecture, so the many security alerts for HP-UX just don't have MPE/iX counterparts. A security breach is never impossible, but the 3000 comes closer to being safe by design instead of protected by patch.

HP support does offer experienced engineers and lab-level advice, but these are values that a third party could equal, given the right personnel. Former HP lab engineers are already at work in many third party companies, some supplying support. For a system like the 3000, which is pretty much frozen in time at HP, there's not very much to keep abreast of that would be impossible outside HP's labs.

Corner cases and corporate requirements carry much of the HP support value in the days to come. The right sized customer will be able to have patches created, once a services contract for the project can be worked out, so a corner case doesn't crater a corporation. And those corporations that demand that the HP badge appear on a mission-critical server's support agreement? They will be needing HP's support value as long as the vendor is willing to sell it.

But while the time rolls back this weekend on the world's clocks, the 3000 community should be looking for the commonplace, everyday value in patches for the system. HP seems to believe that patches are a key value in support. OpenMPE's efforts to assume patch operations looks like it backs up that HP belief. 3000 customers who look at their clocks on Monday morning might want to recall how long it's been since any MPE/iX patch has been so crucial.