Analysis by Rex Barker
Throughout its existence, the HP 3000 has been both adored and abhorred. This love/hate phenomena has historically been polarized into two camps: the people familiar with it (namely Hewlett-Packard and the customers who depend on it), and the rest of the world (who either have some other hardware preference, or are totally ignorant with regard to its capabilities).
It is a natural tendency for human beings to have strong brand loyalties. Nowhere is that more evident than in computer platforms. While the phrase "try it, you'll like it" definitely applies to the HP 3000, it is much more difficult to "try" than sipping a soda. It is also difficult and expensive to "test drive" an expensive minicomputer platform, so overcoming brand loyalty remains a constant battle for the HP 3000.
Let me say that I am very excited about the present state of the HP 3000, and I am equally excited by HP management's genuine interest in the future of the platform. My prediction: the HP 3000 is far from dead. It may have been sleeping, but it is about to wake up and make some serious noise, and get the recognition it deserves for being the best minicomputer platform ever made.
Should I develop on the HP 3000?
The thrust of my comments are focused on the HP 3000 as a platform, and not so much on the various application development tools available, or the best way to build a development team using the HP 3000. Unless the 3000 is a viable and attractive customer solution, there is little reason to develop on it.
My company employs 10-12 developers: we are a medium-sized company struggling to grow, and we are very market driven. We have learned lessons in the past about developing something, and once you're finished, trying to find someone to buy it. We have also spent the past 12 years trying to sell the HP 3000 in a very close-minded world. We've met with some successes and some disappointment.
There are both technical and economic issues that must be factored into any discussion of the 3000 as a development platform. Not knowing where to start, I sat down with a blank sheet of paper and, after some doodling, quickly addressed the following topics.
Reasons to develop & deploy
applications on the HP 3000
Technical issues: 1. Simple, reliable, robust, broad product family -- the HP 3000 song. 2. Excellent OLTP. 3. Networkable, open to interconnectivity. 4. Predictable project management environment.
Economic issues: 1. Keeps application developers focused on applications, not operating systems, networks, hardware. 2. Predictable performance makes delivering what you promised much easier (where I'm from mission critical means this system can't go down). 3. Removing down-time from your developers (and your customers and customer-service overhead) has a economic impact.
Reasons not to develop
& deploy applications
Technical assumptions: 1. Tool vendors aren't jumping to put the HP 3000 on the top of their list of platforms. 2. Job market for HP 3000 skills is not as large as other platforms (I'm not sure if that is true...is the HP 3000 a hard platform for a new employee to pick up?) 3. I need to deliver GUI, object-oriented functionality that is not available on this platform.
Economic assumptions: 1. I can't convince people to buy this platform, so why develop on it? 2. I'm unsure about the future of this platform (if HP can't commit to it, why should I?) 3. Developing on this platform is a one-way street: I can't take what I've done and move it anywhere else.
Reasons for HP to continue
investment in the HP 3000
Technical issues: 1. They have a tremendous pool of talent to bring to bear (the same reason they ported their business applications to HP 3000, not HP-UX). 2. The HP 3000 platform has shown the ability to absorb new technology and direction without crumbling (Classic MPE V to MPE/ix, POSIX, Netware/ix, mirrored systems, multiple processors, WAN mirroring, Internet web servers) and without sacrificing performance and reliability. 3. The HP 3000 is built using the same tools (hardware & operating system code development) as HP-UX. It is not presently on a divergent course. 4. Overcoming MPE's internal limitations that prevent future growth and expansion (using existing tools and leveraging existing knowledge base) may well prove easier than a "shot in the dark" at the next technology platform (beyond Unix) with new tools and staff.
Economic issues: 1. It still remains HP's largest profit stream. 2. Reliability, performance and mission critical dependability is still marketable -- in the right package. 3. Other factors aside (corporate policy, application availability), once an HP 3000 customer, always an HP 3000 customer. That long-term commitment has an economic impact.
Reasons for HP not
to continue investment
Technical speculations: 1. The underlying architecture does not allow for any more future development. 2. The next evolution of the HP 3000 will not allow for complete backward compatibility. 3. The compromises required to integrate MPE with other technologies (Unix, Netware, NT, etc.) pose a threat to MPE's reliability and performance that can't be overcome.
Economic assumptions: 1. We can't convince people to buy this platform (so why invest in it?) 2. It has reached the end of its usefulness: from this point on, it is too hard, or too expensive to enhance. 3. Today's market is reactive, not proactive. We don't know what to do to the HP 3000 to make it marketable in today's environment.
Those were the basic ideas that factored into my company's decision to continue to work with the HP 3000. And I feel certain that HP battles with these same issues. And no matter how technically superior the HP 3000 is, unless HP can make money with it, they can't commit long-term resources without a clear indication of long-term revenue to match.
But don't worry. I'm gonna tell everyone what to do.
Rex's sure-fire plan
for the future of the 3000
Here are the technical and marketing efforts that HP could apply to the HP 3000 to invigorate it as a competitor in the high-end OLTP server market.
Technical efforts: 1) Continue R&D on the HP 3000 (You knew I would say that!). Bring it the processors, peripherals, compilers and operating system functionality that keep it moving forward. If that means scrapping the underlying architecture and starting over, do it. 2) Close the gap between MPE and HP-UX. Make them similar in price performance and functionality. Run one as a shell of the other in the same box. Give them the same "openness. They are now cousins, make them sisters. 3) Deliver IMAGE/SQL on the HP-UX boxes. Give it away to HP-UX platforms, and sell it to other Unix platforms. Use the Netscape, Java approach to attract people to its robust OLTP. This does not threaten the HP 3000, it enhances it. This is the single biggest effort toward bringing the two platforms together.
There's much more. 4) Make MPE the springboard for new technologies. HP hedged their bet in the mid-'80s by using the same RISC hardware architecture and operating system kernels to develop both the 3000s and 9000s, both MPE and HP-UX. Once MPE and HP-UX exist in the same box, since MPE is the more reliable, better performer, make it the base from which NETWARE, NT, and other software evolves. If they become the defacto standard, the HP 3000 is right there, and is smoothly integrated: if not, the investment is not wasted. 5) Encourage RDBMS vendors to move the HP 3000 to the front of their food chain. Publish performance benchmarks that prove the 3000's performance with each tool as compared to other hardware solutions. Make sure new product releases are deployed quickly on the 3000. 6) Attract tool vendors to the 3000. Encourage them to develop shrink-wrap server deployment on the 3000 platform. Help to make it the easiest platform to deploy. Assist them as well.
Marketing efforts: 1) Once those technical items have been accomplished, position the 3000 as a better performing, more mission critical, no-risk alternative to the HP-UX (or any other hardware platform). Don't apologize for MPE -- explain it. The backward compatibility that MPE has absorbed since 1972 has not prevented it from moving forward. Tell the world.
2) Deliver the HP 3000 as a shrink-wrapped, preconfigured application development server. Bundle it with a menu choice of tools and RDBMS and make sure it is attractively priced. Yes, make it dirt cheap to the developers (IBM offered to give us an RS/6000 free for a year). Once developers get a taste of the 3000, they will move toward deploying their products on 3000, and those systems could have more margin built in.
3) Position HP as a company so committed to "open systems" that we have made our entire product offering open (9000, 3000, HP-UX, MPE, even IMAGE/SQL). That would not have worked in 1985, but it will in 1997. You did that on the HPWorld expo floor. Do it in your advertising.
4) Build a complete company thrust around what you have done to the 3000 platform. You are currently a leader in delivering technology. You will retain that leadership position only if you act like a leader. Show the world what you do to every platform you develop. Make it a slogan: Let us show you the HP way!
What we really ask
from HP as partners
Communicate with us. If you don't know what to do, ask some of us to help you form a vision for the future.
If you have a vision for the future, and it does not include the 3000, tell us. Freeze the functionality of MPE, move it over as a shell under HP-UX, and it becomes the "compatibility mode" of the HP-UX box. That will put an end to speculation and we can all draft a plan for our futures.
If there are valid reasons why the functionality of HP-UX is superior to MPE, then please explain that to us.
If your plan for the future platforms is to create a separate product each time those platforms come along, then explain that as well.
We still think of our relationship with HP as a partnership. With all the talk of computer hardware becoming a commodity, perhaps our relationship needs redefining. But until told otherwise, I will think and act toward HP as a close partner, not an adversary.
Customer loyalty is hard to build, but if built properly, is even harder to destroy. The fierce loyalty of the 3000 customer base is a direct result of years of effort on the part of HP. You built the company on customers like us. You made us the way we are. If we need to change, tell us. Give us the tough love, if necessary. But direct, open communication is what we want. You created us. We won't go away, and we won't shut up, either. That, my friends, is the HP way.
Rex Barker is Vice President of Access Data, Inc., an HP 3000 business solutions provider and channel partner since 1982. Access Data delivers multi-entity financial solutions, including standard accounting modules, wholesale distribution, and receivables recovery. Barker leads a team of 11 people that support customers from coast to coast, using the HP 3000 as their primary development platform.