If the wags are right, the combination of Informix and object/relational maven Illustra is Illformix. But even so, is UniSQL too late to even get on board, asks Gary Flood from our sister publication Software Futures.
Chances are that if you’ve heard of UniSQL at all, it’s been through mentions in deals: UniSQL providing technology hither and thither, not UniSQL coming out and saying ya boo sucks to the big bad boys in relational (or even their weirdo cousins in object) databases. But it’s a very good bet you’ve heard of Illustra, even though UniSQL was way ahead of Michael Stonebraker’s outfit, in terms of chronology if not technology (well regarded as UniSQL is in that aspect, mind). Now UniSQL wants to change all that – but is it too late?
UniSQL is a 100-plus company with headquarters in Austin, Texas, the brainchild ultimately of Dr Won Kim, whose work in databases stretches back to the early roots of DB/2 in IBM’s Almaden Labs. Kim knows both relational and object database technology inside out, and rather like rival Stonebraker he had the smart idea of combining the two, not through cramming features of one into the other but by writing a true hybrid product. So even before Stonebraker left Berkeley for the second time to found Illustra (nee Montage) in August 1992, Kim had set up UniSQL (May 1990). In June the following year he got the now-famous $10m seed money to really get UniSQL going through a deal with the systems integration arm of Nippon Telegraph & Telephone. He was thus able to get a commercial product (UniSQL/X 1.0) out in March of 1992. In September of that year he was able to cut a deal with POSC, (the Petrotechnical Open Software Corporation), the user initiative to define open standards for the oil industry that subsequently seems to have faded to black. In August of 1993 UniSQL released UniSQL/M, its multidatabase second product companion to the ORDBMS, which allows access to relational, pre-relational and flat file data sources.
After the NTT link was set up its two most significant agreements were with Cincom (its ORDBMS is part of the latter’s Total Framework next generation offering) and pure-play object database rival Versant, which licensed UniSQL/M to give its database an SQL/ODBC interface last year. These contracts – and POSC – seem to have been the main source of its funding since the NTT deal, since the company says it has received no VC money or other backing (though it adds that it is now seeking such investment to fuel the intro of some new products and for a future IPO). But it has certainly not been spending its money on gaining visibility – in sharp contrast to Illustra, which was able to gain such mindshare that the company was able to sell it self to Informix for Big Wonga. One is struck that the big problem facing UniSQL is that as far as the market seems concerned, the whole object/relational debate is last year’s fish wrapper. Given that Oracle, Sybase, IBM and of course Informix have all put markers in the ground about their object/relational plans (and so has CA, if only to say you can’t do ORDBMS and that pure objects are the way forward), what else is there to say? UniSQL can only kind of hop up and down and cry not fair. Kim acknowledges that Illustra (and now Informix) has spent money achieving awareness that dwarfs his, but claims that that’s all set to change. There will be a stepwise increase in visibility and marketing. As for Informix, they have proven time and again to be masters of hyperbole – all those years going against Oracle has forced them into this Darwinian position just to survive. We can’t match Informix dollar for dollar in marketing, OK – but then Ford put a lot of money into pushing the Edsel.
While none of us were thinking about UniSQL, it wasn’t just standing still. For those of you who have signed up for the Wall Street Journal on the Web, UniSQL is the database behind the Interactive Edition. In June it was able to announce a $10m deal with Korea Telecom, to provide the object/relational core component of an integrated customer information systems project to be rolled out to forty regional telephone service centers. In fact, Asia is a power base for the company – it claims 80 customers (out of 600 users worldwide) in Korea alone. In fact, UniSQL has an extremely impressive customer roster – Mitsubishi Research Institute, Sony, and Toshiba (Japan), Alcatel Bell Telephone, British Aerospace, Elf Acquitaine (in Europe), AT&T, the US military (Army, Navy, Air Force), Westinghouse, Dow Jones, New York Stock Exchange and Lehman Brothers (all US). But… it turns out that when one wants to find out what a lot of them are doing, access gets difficult. Kim promised hassle-free access to some of his great Asian customers: no show. Then we were told that since the company has done no real marketing work at all these past two years, why are we surprised that so few end users have been press-enabled? Then there’s the good old competitive advantage blah. So while customers may be given access to real reference sites, we at Software Futures have no real way of knowing if these cool names are doing anything more than using evaluation copy CD-ROMs as beer coasters. In the early stages, most applications [of UniSQL] were of an engineering nature, says Kim. Recently, though, many are MIS. About 10 percent of our sites are currently beyond prototyping.
WHAT DO THE USERS THINK OF THE TECHNOLOGY?
After some digging in of heels, we were finally given two contacts: one of whom is from commerce, the other is from the hallowed halls of academia. We would, with the greatest friendliness, suggest that Informix-Illustra is somewhat more advanced in terms of leveraging their object/relational customer success stories at this current time, and that UniSQL needs pronto repair work on this side of their story.Anyhow: Ed Frost is software manager at Lockheed Martin, where he is leading a team working on the AIRS (Atmospheric Infrared Sounder) component of the Earth Observing System satellite (a large scale NASA project for eventually multi-petabyte data storage which Sybase and Informix/Illustra both variously claim to have won bits of). This instrument, he explains, will analyze the chemical content and temperature of air as the satellite orbits the earth, hopefully leading to the next generation of weather sensing systems. AIRS is using UniSQL for the real time storage and retrieval of what may become 500Gbs of mainly scientific and engineering test data, needed to be accessed by up to 40 end users.We liked the object-oriented concept of UniSQL but also liked how it could work with relational and SQL, he told Software Futures
Five other systems were evaluated, including Illustra, before the choice was made to partner with UniSQL. We picked it over Illustra because the technology was better; this is a new product, not a makeover of Postgres. But the project is still at far too early a stage to begin talking effectively about how the software is delivering, he notes. However, over the approximately 18 months of evaluation he’s been through, he lauds the technical and post-sales support he has received from the vendor. When we asked Frost about his views on UniSQL as a potential force in the database field, he had a short but effective answer: UniSQL is more of an engineering than a marketing company at the moment. That’s not to be read as a bad thing in his book – but that seems like a fair comment about their approach to the market up until now at least. Even if he hasn’t been using the technology for long, Frost is at least in the commercial sector. The proffered academic reference site was originally Chuck Eastman, professor in Colleges of Architecture and Computer Science at the Georgia Institute in Atlanta. Eastman is leading a team using UniSQL as the database engine upon which they’ve built an engineering database application called Product Modeling. This is still in the prototype stage, but the idea is that it could evolve to a system w here companies from a range of industries could use it to model products (ie an airplane) and the information needed to fabricate them. We found [UniSQL] attractive for many reasons; it supports SQL, it is easy to learn, and you also get most of the object-oriented features you need, he says. Apart from that it also offers some unusual features, such as a very good API interface, sets and sequence structure support, and is all-round very robust and solid. Such functionality has made it relatively straightforward to write a graphical and textual language for the project on top of UniSQL. How has UniSQL been shaping up as a partner? In general they’ve been up to scratch, though we have had some glitches with the new NT version, he says. But he notes that visibility is hardly one of the firs t things that comes to mind when thinking about Kim’s operation. We saw them as practically an underground operation for quite a few years, he laughs. It’s interesting that Illustra came along after them and has made a much bigger splash. I don’t think UniSQL have tooted their horn as much as they could have.
CONCLUSION: IS UNISQL GONNA SET THE WORLD ON FIRE?
UniSQL seems to have good technology, good customers, and a sudden urge to get noticed. But is that enough in the Darwinian (Kim’s word!) world of databases, where technology is notoriously second to marketing sang froid? It depends on what Kim wants to achieve. One suspects that he was quite happy where he was until Stonebraker changed the rules by selling out to Informix and accelerating the whole object/relational take-up.One further wonders what significant added value beyond mere independence he can offer the market? UniSQL must quickly drum up as good a marketing story together as a technology one, or Kim, sadly, will find his rainforest bulldozed down around him.
