Data migration integrity can be in the garage
The Migration Dilemma

Why and When to Leave Platform-Land

Goodbye Tin Man
Life in the 3000 community revolved around platforms. We used to think about these as operating systems. Long ago it was time to change that thinking and call the combo of servers and the OS a platform. You could think of that era as the Land of Oz, instead of OS. It might be time for 3000 owners to change their thinking about computers as platforms. It depends on what else is doing service in your datacenter.

For a small percentage of 3000 owners, the servers built by HP are all that runs in what we once called the computer room. They live in what one storage vendor, one who knows the 3000 well, calls platform-land. Everything in platform-land is connected to a 3000, so the homogenous benefits of multi-server storage just aren't needed.

Companies that live fully in platform-land are using HP-branded devices built exclusively for the 3000. That's the way HP used to qualify its peripherals: tested for MPE/iX. For a while during the years after HP's "we're outta here" announcement, the vendor asserted that any other storage device was risky business. We covered those debates. The results showed the risks were not substantial.

HP's outta-here movement caused movement in the 3000 community, of course. Some of the movement was inbound instead of an exodus. Companies have turned to using Linux servers, more Windows Servers (2008 and later) and even some Unix boxes from HP, Sun, or IBM. That's the moment when a company starts to leave platform-land. You should leave it once you've got multiple OS servers and need to leverage networks and peripherals across all servers. That's the When.

The Why is a little more complicated. 3000s and the Stromasys servers that have replaced the MPE/iX hosts are cradles for the applications that companies don't want to drop. The companies shouldn't drop an application just because it's on the wrong platform. Applications need to exit when they don't serve the business logic anymore. Leaving platform-land supports the continued service from MPE/iX apps. Like I said, it's complicated.

What we know about managing IT in the year 2018 is that it's complex in a way we couldn't imagine in the 1980s and 1990s. The diversity of devices is what's changed the community for good. Even HP began to understand that a multiple-platform storage device was a better value to anyone who didn't live in platform-land.

For almost all of the HP 3000 customers left today, their server isn't the only box in the shop. They have added Windows, Linux and Unix, yes. More often, those operating systems aren't part of the decision loop. We always said that it was the applications, stupid, when deciding what computer was going to get the assignment. The 3000 survives because of its apps, but its survival is beset by more than the calendar pages slipping away. The differences in connected devices chip away at the 3000's rightful place. 

Oh. Not only do we have to try to retain MPE/iX expertise, we also have to keep that Jamaica disk set running. A set that does nothing else for everything else in the datacenter. When it fails, we have to find a replacement for that device with moving components. An old replacement, at that. If we'd made the 3000 ready for a platform-agnostic environment, so it could use modern storage like the later-generation XP arrays, we could justify a replacement and get something newer.

Now the 3000 goes onto the bubble again. The most-quoted reason for companies to scrap their MPE environments, their platform, is because of aging hardware and the worry over replacing it. Peripherals like storage and tape now have their own worry points, according to support companies serving 3000s. Tape is a crapshoot. People have been warning about old disks for a long time.

The Why to leave platform-land is all about the future. When a datacenter is no longer thinking about 3000-only hardware -- when that hardware can be commodity high-class Intel servers or an array that's beyond the reach of platform-land -- the apps can stay off the bubble. The need to migrate recedes, so long as there's expertise about MPE/iX somewhere.

MPE know-how isn't automatic anymore, but there are support companies that specialize in it. The future of self-maintaining that know-how isn't bright. Indie companies can help, and at least the devices need to be useful to all hardware. Until that's done, being stuck in platform-land is just one more way to hurry to an exit from everything still working from the 3000 world, built and configured in the era when platforms were like the heroes from Oz. We'll miss you most, Tin Man.