NewsWire Q&A: Ken Sletten

Getting Customer Development Needs on Target

Ken Sletten has a great bearing on putting customer advocacy on target. As the current chairman of the SIGIMAGE special interest group, he's managed the proposal and balloting process that's yielding significant enhancements such as b-tree indices for IMAGE/SQL. The system manager works at the Naval Undersea Warfare Center Division in Keyport, Wash., managing an in-house software suite that performs all shop processing and engineering control functions for the US Navy's Mark 46 and Mark 50 Lightweight Torpedo programs. He took a bead on his first advocacy target for the HP 3000 in the early '90s, when he was one of a handful of users asking HP to start new development for Transact, a then-flagging language. SIGRAPID members, which included Sletten, wrote the first pages of customer communication with HP at a time when "Customer First" was just a wish instead of an HP 3000 division directive.

Sletten's experience with the 3000 dates back 15 years, when he became system manager for a Series 42 HP 3000 and later a software engineering team member. After serving as the SIGRAPID chair for two years, and as IPROF chair since its first incarnation by the Interex SIG Coordinating Committee, Sletten has had lots of time to talk with HP management about maintaining the value of the 3000. This spring he made a proposal to make that happen, routing support funding to the R&D arm of the 3000 division. HP 3000 division manager Harry Sterling's first reaction was to suggest an alternative, one that would have customers funding special projects on their own. A customer with a lot of ideas and access to HP, Sletten has a clear set of targets about HP's 3000 goals. Since Sterling's direct-report manager changed recently, we believe the thoughts of a customer like Sletten would make a great primer for the new chief of the Enterprise Server Group on what could be done for MPE customers. In the second of a two-part interview, Sletten gives a detailed look from the SIGIMAGE chairman's seat at what HP needs to do to maintain the value in IMAGE.

What is the most pressing need, in your opinion, for HP's database products?

R&D funding needs to be increased so that the most important enhancements in a long list of known worthwhile improvements to IMAGE and Allbase can be implemented in a timely fashion.

Which of the features now on the drawing board now should HP put its database development resources into for earliest delivery?

Since bundled 32-bit ODBC is finally well beyond the drawing board stage, I'm going to skip over that obvious long overdue need. And sorted sequential access, a.k.a. IMAGE intrinsic b-trees, is now is beta test, so that is almost out the door. Both of these pending enhancements will provide major additional functionality and flexibility to users of Image/SQL.

That leaves ODBC access direct to TurboIMAGE, MPE, and KSAM files, without having to go through Allbase/SQL, as the overwhelming winners on the 1997 SIGIMAGE ballot. But these features are available now from two reputable third-party vendors. Plus HP has made it quite clear that they do not intend to compete with the third-parties that are offering these extended ODBC features.

Given where we are right now and the length of time it took to bundle the ODBC via-Allbase feature, my personal opinion is that IMAGE users would be better off to have HP spend its R&D resources doing some of the other IMAGE "internal" enhancements. Note that I want to make it clear that I am speaking strictly for myself on this issue, not as SIGIMAGE chair.

In any case, when you get into specific enhancements you are of course in a site specific, your mileage may vary situation. My answer might be a bit different depending on whether I put on my customer hat or my SIGIMAGE chair hat. But generally I would say that the enhancements that have bubbled to the top of the SIGIMAGE ballot over the last couple years are probably a pretty good place to start; in addition to the ever-present and ongoing scaleability effort: date/time datatypes, support for Binary Large Objects and the ability to add/drop datasets, indexes and items through DDL.

How can HP get around the problem of limited staffing in its database section when the best qualified people aren't available to work by the hour?

Harry Sterling already had a good answer for that one: Contract out the work to some of the technical experts in the HP 3000 third-party community, funded by input of additional user dollars.

It feels like it's been a good long while since a major enhancement to IMAGE has surfaced from the database lab -- jumbo datasets really sticks up on the horizon, some two years ago. What's your feeling about the rate of delivery of things like ODBC and b-tree indices?

HP was clearly way behind the power curve with respect to the time it took to deliver basic bundled 32-bit ODBC capability. B-trees also took longer than we would have liked. Beyond these two specific cases, there were 48 items on the SIGIMAGE ballot in 1997. By the time the SIEC gets done adding new requests to the 1998 enhancement ballot that we expect to have ready by HP World - Chicago, I expect we will have over 55 items on the "to do" list even though a couple dropped off. All this speaks loudly to the need for an increase in 3000 R&D funding.

HP likes to hold out Oracle as an avenue of applications for HP 3000s, but at the moment it's not certain the latest version of Oracle will be ported to MPE/iX. (It's not doubtful, either.) Does this Oracle "net" make you feel any safer as an HP 3000 manager?

Putting on my customer hat, from our site's perspective the answer is no, since we fully expect never to run a DBMS other than IMAGE/SQL on our HP 3000; other than the very general fact that every additional customer on a 3000 is good for the platform. But given Oracle's position as the leading portable DBMS vendor, the fact that it runs on the 3000 and will support gateway access to IMAGE/SQL is clearly a consideration for sites that deal with Oracle environments.

HP seems to believe that the enhancements proposed by SIGIMAGE are being requested by a very small percentage of its installed customer base. What can you say about something like time/date data types when it appears on less than 100 customer requests in a poll over the Internet and through the Interex membership?

I believe quite a few sites look to SIGIMAGE for leadership in working with HP to get enhancements incorporated that will keep IMAGE/SQL moving forward and competitive in the current computing environment -- even if they do not actively participate in SIGIMAGE meetings or balloting. I have had users from various sites tell me this on several occasions. But I will admit, I wish we could get a larger return on the yearly SIGIMAGE ballot. Quantity is not the only measure, but it certainly is one measure that HP looks at. We are considering some ideas to try and improve the number of sites that participate in the process.

What problems do you think might happen if DDX doesn't get altered to work with datasets greater than 4Gb?

Taking somewhat of a long-term view, this is tied to the question of how long it will take HP to actually deliver a 64-bit file system for MPE, and then to when and how IMAGE/SQL will make use of it. In the meantime, sites that run 7x24 or close to it with very large databases are impacted by this restriction. But I don't have any numbers on how many if any sites consider this a major problem in the near term. If there are IMAGE users out there that really need this now, I hope they will contact HP or SIGIMAGE and let us know. I can be reached at:

Your question has just prompted me to add another enhancement request to the SIGIMAGE ballot for 1998. We'll give the users a chance to put in to a vote.

As a customer, do you trust the work HP might contract for with outside programmers to enhance the operating system and the HP databases?

At an open meeting at IPROF-97 Harry Sterling said: "I have complete confidence in Allegro" (Consultants, Inc.). I am in full agreement with Mr. Sterling on this one. Note that although Harry only specifically mentioned Allegro, there are perhaps a few other firms with long experience and a proven track record on the HP 3000 that might also be considered for this type of contracting out. A couple come to mind, but I'll leave HP to say what they think about that.

If HP is going to defer to third party solutions -- either delivered or promised -- as the answer to problems like ODBC access, what's the point of having a database enhancement request ballot?

This is an ongoing process, subject to change and modification if enough users continue to request a particular feature or product from HP. It also makes a big difference whether or not the item in question is an external access method like ODBC, or an IMAGE internal enhancement that only HP (or an HP direct contractor) can do.

Should HP do everything it can on the ballot, regardless of third-party offerings or promised products?

In my shortest answer so far, no.

What is it about IMAGE that's given the database its ability to withstand the allure of relational databases?

Performance, reliability, performance, simplicity and performance. And maybe the fact that it is thoroughly understood at a detailed technical level by a couple of thousand people around the world, both inside and outside of HP. The popular and "portable" third-party RDBMS systems are pretty much black boxes to anyone outside the company that sells them -- in some ways they are a lot more proprietary than IMAGE. But they are portable. Of course if you stay with an HP 3000, that's not an issue.

What's changed about CSY's relationship with its customers since the the first tangible SIG success with SIGRAPID, an effort that got HP to rejuvenate Rapid products for the 3000?

Bob Meissner gets the credit for initiating SIGRAPID and starting the ball rolling on Transact enhancements. My current fellow members of the SIGRAPID Executive Committee Cecile Chi and Nick Demos have been with SIGRAPID from the start.

Since 1990 CSY has progressively become more open and user focused, as demonstrated by interactions with SIGIMAGE, SIGMPE, SIGRAPID, and other Interex SIGs. The biggest changes on the HP side have come outside CSY. During the regime of Wim Roelands, it looked for quite awhile like all HP wanted to focus on was HP-UX. Especially since HP World - Anaheim in 1996, there has been some noticeable movement to a more even-handed treatment of all of HP's OS offerings: MPE, HP-UX, and now of course NT. But still, in the end money and how much there is of it drives everything.

As a customer, what kind of leadership do you expect out of a Solution Team manager handling HP-only things like Rapid and database development?

To make realistic projections of what HP can and will do with HP-only products within the limits of given funding parameters, and then do his or her best to deliver on them. Also, to as much as possible take a proactive approach with respect to changing user needs that may be driven by external forces; i.e.: don't necessarily wait for everything to bubble to the top of the user enhancement request list, if there are issues that obviously need to be addressed. For example, IMAGE QUERY support of IMAGE b trees. We in the SIGIMAGE Executive Committee all forgot to put it on the 1997 ballot -- maybe we all subconsciously figured it was so obvious that we didn't need to.

Harry Sterling says he's not in the database business. How does that make you feel?

I can't say it leaves me shocked and surprised, since HP CSY management going back to Glenn Osaka has publicly said the same thing for years in various user-group forums. And in one sense it is obvious that HP is not and never will be a "pure" database software company like Oracle, et. al.

Be that as it may, I wish HP management would try and find a better way to state their overall focus in a way does not appear to deliberately downplay what I believe is a fundamental cornerstone to the past, present and hopefully long-time future success of the 3000: IMAGE/SQL (which by current definition includes eight out of the nine software modules that make up Allbase/SQL). I realize that there are a number of HP 3000's running production systems that use neither IMAGE or Allbase, but I would bet that in the current world-wide population of online machines, the overwhelming majority are still using IMAGE for most if not all of their production operations. They may be adding on client-server extensions like we are, but IMAGE/SQL is still at the core.

Therefore I believe the "triad" view of the 3000 is still valid; i.e.: At the core of the 3000's success is the tight integration and system-level tuning of very reliable hardware, the MPE OS and IMAGE/SQL. Take away any of these three legs, or have them start to fade or fall behind to any degree, and the 3000 will be in trouble. Selling a DBMS may not be HP's primary business, but I would respectfully maintain that as good as MPE/iX is (and it is a very good OS), without IMAGE the HP 3000 would have faded from the scene a long time ago.

Therefore even if HP is not in the "database business" per se, they are certainly engaged in the business of selling a system that is critically dependent on the IMAGE/SQL DBMS.

Copyright 1997, The 3000 NewsWire. All rights reserved.