September 2003

Build, Buy, Port or Stay?

What an IT organization committed to the 3000 should do

Port: Pros and Cons

This is the option getting the most attention from HP and its partners, for obvious reasons. It is potentially the fastest, particularly if you load up on middleware translation tools and do an emulation port. Doing an emulation port also reduces training costs. If you take more time (and possibly spend more money initially) to do a native port, then this can give you a fairly attractive TCO over time compared to the emulation port because you eliminate the ongoing license fees for the middleware tools. It also lets you take better advantage of emerging technologies and achieve better overall performance than if you do an emulated port.

Porting is particularly well suited to a staged approach over time. If you can use Eloquence as your target database, you receive significant savings since it is far less expensive than the other commercial DBMSs. Eloquence also natively emulates TurboIMAGE so that, in most cases, you will not need to change either the database logic in your code or even the calling code itself.

On the negative side, a “port” necessitates training and recurring costs for migration utilities. A port to an emulation environment may just be “postponing the inevitable” and only serve to increase your TCO and decrease your ROI. It may introduce performance issues as you try fitting logic tuned for TurboIMAGE into a relational database. A port project may require an extensive period of “code freeze.”

Ported marginal code is still going to be marginal code. If you have a mixture of 3GL and 4GL code, the porting tools may not work as well as you hoped. Of all the options, porting is the hardest to show its ROI. Initially, HP and its partners advised against doing any development on the HP 3000 during the port because of the obvious complications it introduces. However, because of the ROI issue, many are now suggesting that you use the “opportunity” to either add or improve functionality during the port.

Stay: Pros and Cons

“Stay” has a lot of obvious appeal for as long as support and parts are available because there is no need to learn a new platform. Furthermore, staying is also clearly the least cost solution short term (through 2006), even if you buy a new system(s) sooner than you originally planned to get in under the October 31, 2003 end of sales date. Staying might be a good first stage option in a two-stage transition strategy where the first stage is staying on the HP 3000 for “n” years while continuously evaluating options for the second stage starting after n years. Most other options require you to become a systems integrator. You buy the basic platform and OS from one vendor, a DBMS from another, a scheduler from a third, a spooler from a fourth, backup software from a fifth, etc. Managing all these different vendors and ensuring that everything works together can become a nightmare.

Staying is not without its risks and downsides, however. In my opinion, support, unless you are in a major metropolitan area, is problematic for the period 2007-2010. After 2010, regardless of where you are located, support and parts are likely to be spotty and expensive.

You need to ask yourself whether by staying you are just postponing the inevitable. If you rely on third-party vendors for key software, what will their status be in the outlying years? What about the people who wrote and maintained your applications — how long will they be around? There is no new talent being trained on the HP 3000. Right now, there is a glut of HP 3000 talent available, but five years from now that pool will have diminished considerably and what is left may well be very expensive.

You also may not have time to update your equipment prior to HP’s October 31 end of sales date. Do not assume you will be able to pick up an inexpensive A- or N-Class system in the next couple of years. You are likely to find they are neither readily available nor inexpensive.

There are several reasons for this. First of all, not that many A- and N-Class systems have been sold to date. Secondly, those systems that have been sold have been sold relatively recently. These systems have gone to customers who either expect to keep them multiple years, or convert them to HP 9000s using HP’s free conversion offer.

Next month: How to go about making outsourcing and a portfolio approach work for you, and some recommendations.