Genisys Sponsor Message

June 2002

The Contradiction of Openness

By Scott Hirsh

en·tro·py (n) ‘en-tr&-pE; Inflected Form(s): plural –pies 1. A process of degradation or running down or a trend to disorder

As The Firesign Theatre once said, “Welcome to the other side.” We’re now in the post end-of-life world of the HP 3000 – we don’t need to pretend to go along with that e3000 nonsense any more – and options are now being weighed. Inevitably, most users will be forced one way or another, for better or worse, to move on to new platforms. Because we’re now at the crossroads of an important and far-ranging decision, it’s time to revisit IT decision making.

While we can’t prevent those we cannot control – think pointy-haired – from making bonehead decisions, we can at least be realistic about the decision making process and brace ourselves for the horrors to come. In other words, if you have nothing good to say about the way important decisions are made by IT management, come sit by me.

As someone who spends a large percentage of his time on the planet with IT decision makers – and having been one for more years than I care to admit – I can state with authority that the customer-vendor relationship boils down to who thinks he has the upper hand. Having the upper hand means you get to control the terms of the deal. And isn’t this the goal of any negotiation?

When discussing this topic I am reminded of an ad by the Post Office I used to see when riding the trolleys in Boston: “If It Seems Too Good to Be True… It Is!” Everyone on the buy side likes to consider himself in control, and vendors do everything they can to pander to the need customers have to be in control. And no message I have seen during my career is more patronizing than “openness.” For the purposes of this column, open is now defined as “not mainframe.” I know it’s not fair, and I know you disagree (how about the AS/400?), but that’s pretty much the way the world sees it. And I won’t fight it for this discussion.

Open is “good,” proprietary (so-called) is “bad.” Open means freedom, proprietary is a prison. To get to the heart of the issue: open is cheaper, proprietary is more expensive. It would take a lot longer than a column to debate total cost of ownership (TCO), but for those paying attention – management, financial people – it’s the total cost of ownership that matters. However, in tough times like now, the discussion stops at acquisition cost, or what it costs to buy the stuff today.

It seems that especially in times of tight budgets, customers revert to anywhere from suspicion to outright hostility whenever a vendor brings up TCO. Nevertheless, there is more to the cost of technology than how much its hardware and software costs. There are hard costs like maintenance and upgrades. And there are those twitchy, but nonetheless real, soft costs, like integration with other technologies, and potential downtime. And then there are people.

The HP 3000 could be supported by average people — you and I are special, everyone else is average — with high availability results. I personally hired non-technical people off the street and had them productive in HP 3000 operations in days (easy to learn, easy to operate). MPE was designed to halt rather than corrupt data. Everything about the HP 3000 screamed low TCO. But we lost, and now we’re hearing that we’ll be better off on open systems anyway, because that proprietary stuff is more expensive.

The reality of the situation is that acquisition cost was higher for an HP 3000 than, say, an HP 9000. But after that, every aspect of HP 3000 TCO is cheaper. Sure, HP’s decision was based on the size of the installed base it could still see, and the lack of big name applications based on big name database management systems. Nevertheless, if anyone’s management thinks that moving from the HP 3000 to something else is all for the best anyway, think again.

You, my friend, are now more valuable than ever. The more “open” the solution – like Linux – the more critical the IT people become who support it. Isn’t it ironic that with all this economizing, people are more important than ever to the success of your environment in its next life? Yes, you must now acquire new skills – ka-ching! – or consultants and contractors must be engaged – ka-ching – or new staff must be hired – a little cheaper, perhaps, but do they know your company’s business (ka-ching!)? And everything you already know – e.g., change control — is even more valuable in the less structured open systems culture.

Okay, so we concede that open TCO is really higher after all. But it’s open, so I can just move to another hardware platform if my vendor forgets his place and treats me like I’m trapped in a proprietary world. Wrong! Can you move with binary compatibility from HP-UX to Solaris? From Solaris to AIX? To AIX to Linux? Even from one flavor of Linux to another? No, I’m afraid it’s not that open. Heck, you can’t even get support for every platform on everyone’s storage or everyone’s RDBMS. And if you change one thing – HBA driver, OS level/patches, PDC, whatever – you may even break the environment you’re on now. Not to mention that you may need to cluster just to get the same high availability you got from old reliable.

In summary, the truth about open systems is that:

• Vendors have as much control over customers as they did in the proprietary days. That is to say, transfer costs – the costs to switch vendor/platform – are still daunting.

• Acquisition cost (new) is probably lower, but total cost of ownership is at least as high if not higher.

• People are the glue that holds open systems together. Skill level must be greater, learning curves are steeper.

• No vendor wants to be truly open. If there’s an open standard, everyone will create proprietary extensions. True openness means commodity, and no vendor wants that. (That would be like two companies merging so they can be #2 in PCs, a commodity product.) Write once, run anywhere? Sure, as long as doing that kills our competitor so we can then rule.

• All vendors want nothing short of Total World Domination.

Worst Practices in Open Systems Purchasing

Just as it’s more difficult to own open systems, so too is it more difficult to purchase open systems. Consider the following worst practices I observe daily.

1. Not Thinking Long Term. Long term in IT is generally three years. But it still makes good business sense to plan for business relationships that last longer. Changing vendors for mission critical systems, no matter what the platform, will always be painful and expensive. It’s worth the effort to work together to create strategic partnerships.

2. Being Emotional, Not Rational. Despite protests to the contrary, most IT buying decisions are not rational. Rather, they’re part inertia, part personal and a little rational. It’s certainly not based on anything as straightforward as price. Most vendors are so hungry for business that they’ll meet practically any price to gain market share. So emotions are typically exploited. Furthermore, IT people tend to treat technical preferences like religion. Diversity of technology is a fact of life. Nevertheless, we always hear of “Unix bigots,” “Windows bigots,” and so on, because only one technology can be the One True Way.

3. Underestimating the Value of Support. The more diverse your environment, the more difficult it will be to keep it up and running smoothly. You got spoiled by only needing to call one number to get all your system problems resolved.

4. Reluctance to Trust. Having “one throat to choke” can be both a positive as well as a negative. It goes against the whole idea of openness to buy most, if not all, components from one vendor. And even if you do, chances are, that vendor just OEMs some of the components anyway. But there’s a lot to be said for integration or system testing, and having one number to call. This option is too often dismissed out of hand.

5. Not Including All Costs in the TCO. The “T” stands for total, so conveniently leaving out certain costs to back into your pet solution is a classic worst practice that will come back to haunt you if you stick around long enough. A recently publicized danger in this area is to blindly accept a vendor’s TCO estimating tool. The vendor tools are not without merit, but they too have a bias (duh). So use the vendor TCO calculators to jump start your own comprehensive calculation.

6. Not Having a Contingency Plan. Everyone makes mistakes, and even the most methodical, systematic evaluation process can result in a mistake. We already know from experience – which is why we’re so valuable – that we should check references and run a pilot. Even still, if we’re choosing an Oracle application on HP-UX, we may need to adjust it to Solaris or AIX if something goes wrong. Or a different VAR, or even a different application, worst case. It’s emotional, not rational, to want to stick with a technology when it’s proving unworkable. The more flexible you are in the early stages, the more open your options. 7. Going Cheap Where It Doesn’t Pay. Does it make sense to back up your mission-critical data with the included program dump, which is “free?” Probably not. But that is not to say a freebie utility doesn’t make sense in other areas. I see too many situations where a customer decides they have spent “enough” on the hardware and main application, which leaves nothing for other critical tools. Just because the old HP 3000 FOS software worked great doesn’t mean you should run your shop on Gnu tools.

8. Underestimating the Value of People. Those who view open systems technology as a commodity probably view the people who support it the same way. When Unix first invaded my shop, Unix contractors were brought in to help with the transition. After demonstrating 10 years of near-zero downtime, I was amazed to learn from these contractors that I knew nothing about computers, not to mention operations and systems integration. Silly contractors. Command syntax, performance tuning and the like are the easy part. Proper procedures, judgment and knowledge of your company’s business are where the value remains. If you ran a great HP 3000 shop, you’ll run a great shop under any platform.

Close Your Eyes and Think of England

For me, the HP 3000 has been a great computing platform for the last 20 years or so (I know, to some I’m a new guy). Since heterogeneous computing is already a reality everywhere, we’re looking at disappointment more than shock. It’s not a religion (see #2), but it will certainly be a challenge to develop the same kind of relationships we have enjoyed through our association with the platform. The worst practice of all would be to make a sloppy transition, both technically and personally.

So if you’re leaving the HP 3000 for another platform, best wishes for a smooth migration — and I hope we keep in touch. In an open world, the most durable advantage may well be your ability to maintain alliances with those you know and already trust. When the horizon seems to darken, it’s the people that hold everything together.

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 advising on automation and administration.