Worst Practices: Dead Right

May 2001

Dead Right

By Scott Hirsh

When does a virtue become a vice? And when does a best practice become a worst practice? This theme arose in a conversation that began with the news of HP’s plans to discontinue the Legato Networker client. Why is it that the HP 3000 customer base can’t seem to generate critical mass for products like Networker? It can’t be only the size of the HP 3000’s installed base. Plenty of software companies (including HP) are doing just fine selling to HP 3000 customers. No, there must be more to it than sheer numbers.

Clearly, if Legato had achieved a sufficient share of the HP 3000 backup software market, the Networker client would still be with us today. After all, companies like Legato calculate the potential of any given market before committing to it. Sure, one could argue that the port of their client software was inferior and deserved to die. And we do seem to make that argument every time another popular product bombs on the HP 3000. But having sponsored one of the early Legato presentations at a SIGSYSMAN meeting, I can’t help the sense of déjà vu remembering the enthusiasm that accompanied product kickoff. In fact, back when MPE/XL was first being unveiled I organized a SIGSYSMAN meeting with all the database vendors who were going to port to the HP 3000, including Sybase, Oracle and Information Builders. That effort, too, started with great fanfare then fizzled.

Which brings us back to best practices that become worst practices. A common HP 3000 system management best practice is not to jump on the first release of any product. After all, stability is the essence of the HP 3000: stability in terms of no downtime, stability in terms of upward compatibility, and stability in terms of data and investment protection. Always online, that’s us.

But what happens when stability manifests itself in a reluctance — even an aversion — to adopt new products and practices, even after they are widely adopted elsewhere? What happens when preserving our familiar environment in the name of uptime becomes… a rut? And what happens when this rut results in the HP 3000 cementing its place in the computing world as a niche?

There’s nothing fundamentally wrong with a niche market. But there are some trade-offs that must be reconciled. First of all, major software players will not be attracted to the platform. Like it or not, if one of the major database vendors doesn’t support your platform, the applications that are based on that database will also not be available. And it’s applications that sell boxes. And the more boxes that sell, the more likely it is that other applications will be ported to the platform. And so on (rinse, lather, repeat).

Early bad luck with database vendors (particularly Sybase), continuing to the present with Legato, has pretty much established the HP 3000’s position in the overall computing marketplace. Who did or didn’t do what when is not important at this point. What matters is the behavior of HP 3000 users, who tend to buy products that established HP 3000 market share long ago (or just buy HP). We stick with the good old CI and avoid Posix. (If you’re one of the two or three Posix die-hards out there, feel free to flame me). IMAGE: yes; Allbase: no. Command line: yes; GUI: no. We’re creatures of habit who take our service-level responsibilities seriously, even if it means avoiding many of the products and trends widely adopted elsewhere.

I see the inside of a lot of non-HP 3000 datacenters these days. Needless to say in this forum, (so-called) “Open System” often looks more like “open sore.” It’s ugly out there, a world where stability is frequently just a dream. (Don’t be fooled by the calm exterior. Inside, the headless chickens are in full motion.) So maybe we HP 3000 managers are right in many ways. At least I think we are. But the other guys have market share. And market share has its rewards – like variety of products to choose from.

I remember a certain well-known curmudgeonly IMAGE utility developer snorting at the term “legacy system” as applied to the HP 3000, especially in comparison to Unix systems. “Unix is a legacy system!” he exclaimed (and that was a good 10 years ago). His point was that the HP 3000 was being perceived as old and outdated, while even older Unix was somehow “state-of-the-art.”

At some point the truly state-of-the-art HP 3000 developed a perception problem, a problem that doesn’t seem to be as severe with the AS/400, which shares some of the same “legacy” and “proprietary” knocks. But that’s been the easy straw man for why the HP 3000 can’t seem to grow its market share. It’s time we admit that for whatever reason, we haven’t put our money where our mouths are. We don’t want to operate our HP 3000s the way other platforms are operated. We like individual, host-based backup. We like hierarchical databases. We like command line. We like Just a Bunch Of Disks…

All of which are best practices, when you consider our track record as a community of System Managers. But at the same time, this is a business. Third-party vendors and HP will invest most where they think they’ll get the best return, and it won’t be the HP 3000. Our system management best practice, alas, turns out to be a business worst practice. It’s food for thought the next time you submit a product enhancement request, or wonder why you’re being forced to implement a new application on a different platform.

Scott Hirsh (scott@acellc.com) former chairman of the SYSMAN Special Interest Group, is an HP Certified HP 3000 System Manager and founder of Automated Computing Environments (www.acellc.com, 925.962.0346), HP-certified OpenView Consultants who consult on OpenView, Maestro, Sys*Admiral and other general HP e3000 and HP 9000 automation and administration practices.