Unifying Sun's Open Source Software

- by Paul Murphy -

Until quite recently Oracle's salespeople would recommend Sun hardware because SPARC offered the memory, processor speed, and reliability needed to make the database product seem pretty good. Today, however, Oracle sees Lintel (Linux on Intel) as its route to a bigger share of the customer's budget. That's bad news for Sun even if the Linux hardware comes from Sun, because the margins are a bit slimmer and the chances of damage to Sun's reputation as a reliable supplier far greater.

To make things even worse, it looks like Oracle will succeed in killing off Peoplesoft, thereby seriously limiting the customer's alternatives and taking away another Sun marketing channel.

So what's to be done? Hand wringing isn't going to work: the right answer is for Sun to buy Unify corporation and do a StarOffice with it.

Unify has a market cap of around 12 million, some of the least effective sales people you've never met, and great products -including a decrepit relational database engine that blows away the competition, an ancient Unix development environment that's truly great, and a rather buzzy enterprise java development suite that has a more mature focus than Sun's own Java Enterprise Suite.

The Unify 2000 RDBMS (U2K), now sold as Dataserver, isn't the fanciest database manager in the world and lacks many of the advanced features found in Oracle's RDBMS products. On the other hand, it works well with Solaris on large SMP machines, is quite a lot faster than Oracle on similar hardware, and is sufficiently robust that it usually takes an MCSE with the root password at least several days to break it.

What Sun should do with this is similar to what they did with StarOffice - expand the existing organization to directly support and enhance a truly free and open source version for Unix while offering licensing for those who prefer the usual contractual vendor relationship. The result would be an extensively proven, open source, RDBMS ranking behind the full PostGres product set on advanced features, but well ahead of MySQL on performance, scaleability, and readiness for enterprise use.

An open source U2K would give people who want more than MySQL can provide but don't need all the features offered by the Oracle RDBMS the opportunity to make a better choice at a much lower cost - and give Sun a significant selling advantage on the hardware because it takes full advantage of the highly scaleable SMP now available only on higher end Sun gear.

The combination of Unify's older Accell/SQL and Vision products constitute a powerful development environment and should be enhanced as a licensible, but also free and open source, product on the StarOffice model. This toolset is both archaic and state of the art - because it's not a Windows style IDE. Instead Vision automates forms development while Accell/SQL adds programming language support, making the combination state of the art for data driven development directly on Unix -and that's where the next Peoplesoft is going to come from.

In combination, Accell/SQL, the Vision front end, and the U2K data management software complete Sun's open source software architecture - allowing big companies to build and deploy complete open source solutions on Sun hardware from the server room to the spiffy new Sunray desktop.

Accell/SQL and Vision aren't java tools. Unify has those too: their NXJ enterprise development and run-time product set is largely complementary to Sun's own Java Enterprise Suite because it focuses more on deployment and run-time support than on moving Windows developers to Java.

Java is pretty much a requirement for enterprise entree these days but only because people are widely committed to Microsoft's client-server architecture. What's going on is that the stability of the Java run-time machine as a development target makes it extremely valuable as a means of reducing the impact Windows variability and the Microsoft upgrade cycle have on client-server deployment -exactly what also what makes Java so useful for embedding functionality in things like third party cell phones or DVD players where you need to be sure that millions of cheap copies will work exactly correctly each and every time.

On the other hand, if you don't buy into client-server to begin with, you don't need the kludgy Java run-time machine to get around its consequences. In other words, companies using Sunray desktops don't need to incur Java's overheads and can, instead, use the forms based applications model embedded in the Vision Accell/SQL toolset to deliver very high quality applications at a small fraction of the cost and resource consumption involved in delivering more politically correct Java applications in client-server mode.

That doesn't mean Vision won't run or deploy on PCs. It does, in fact the latest versions are somewhat Windows centric and some clock unwinding will need to be done to recover from that because the real power of the toolset shows up only when you get rid of the PC. Together with the Sunray, Accell/SQL and Vision amount to a client-server killer: a click and drool GUI environment that's fully object oriented with re-usable libraries for those who care, but with minimal overheads and genuine usability for those who just want to get the job done.

On the other hand, while I don't believe that Microsoft's client-server architecture has a long term future in business computing, it's certainly dominant now. As a result re-invigorating Accell/SQL with Vision as an open source Sunray development tool is a bet on the future. To also bet on the present, what Sun needs to do is combine the best ideas in Unify's next generation Java environment with those in Sun's own enterprise java suite. Doing that would add maturity to Sun's current product by shifting the design focus forward from the IDE stage of the application lifecycle to more completely include consideration of deployment issues and the messy business of getting along with continuing applications supported outside the Java environment.

Doing all this would cost Sun about twenty million bucks in the first year and five to ten million a year after that. Is it worth it? Well, what would Sun get? Addition of a powerful database and Sunray development toolset to its open source software suite, a much stronger Sunray story for businesses, and the opportunity to tell customers who might otherwise be forced to choose only between IBM and Oracle, that Sun has a better, cheaper, faster, product set to offer.

Kind of a no brainer isn't it?

Paul Murphy wrote and published The Unix Guide to Defenestration. Murphy is a 20-year veteran of the IT consulting industry.