What is Actian actually doing? - Why does Actian need so many databases?

Actian, previously Ingres, is beginning to act like its erstwhile parent CA. Or, at least, how CA was acting back in the 90s, when it was acquiring everything in sight. Back in December, Actian acquired Versant, this month it finalised its acquisition of Pervasive Software and today (April 25^th) it has announced the immediate acquisition of ParAccel.

In other words the company has gone from owning two database management systems (Ingres and Vectorwise) four months ago to now owning five (adding Versant, PSQL – more popularly known, at least to old fogeys like me – as Btrieve, and ParAccel). The question must be why on earth would it want so many?

Of course, PSQL and Ingres are cash cows. Moreover, having Ingres allows Actian to claim that it has tens of thousands of users (even before the Pervasive acquisition) – which it does – even when promoting its Vectorwise product – which doesn’t. And, of course, there’s plenty of synergy between Actian’s database products and Pervasive’s data integration and data governance portfolio.

You can also sort of see why Actian might want Versant: because it wants to exploit its technology in the big data space for unstructured data. And what used to TurboRush for Hive also fits into this picture. In fact, Actian’s web site makes lots of play about solving big data problems, though as far I can see it has a bunch of parts rather than anything integrated yet: there’s is some work to do in pulling everything together but you can see that the potential is there.

But why ParAccel? After all, Actian’s marketing is all about the “fact” that Vectorwise is the “the world’s fastest analytical database”. In which case, why would it need to acquire a (slower) competitor? Of course, the truth is that it is “the world’s fastest non-clustered analytical database, according to TPC-H, for capacities up to 1TB, of those companies that have bothered to do a test” and, as it happens, ParAccel (which is clustered) is faster than Vectorwise according to TPC-H.

So, what has ParAccel going for it? As far as I can see: first-class optimisation technology, a (small) cash cow in Amazon which has embedded ParAccel technology into Amazon Redshift (its cloud-based data warehouse offering) and a not very substantial user base. The marketing blurb on the takeover states that ParAccel is a market leader in the big data space. Well, I suppose it depends on what you call a “leader” but, however good ParAccel is from a technical point of view, my understanding of the word leader is clearly not the same as that of Actian’s marketing department.

When I first heard about this acquisition I figured that there was one of two conclusions to draw: either Actian was having trouble scaling Vectorwise and needed a different platform to move up-market with or Actian sees the acquisition as a way of acquiring some pieces of useful technology plus increasing its market share. In turns out that it is the former that is the case: the company is talking about Vectorwise up to around 20TB and ParAccel thereafter.

Given that this is the case what is the likely future for both Vectorwise and ParAccel? It cannot conceivably be right to have two such products. Either one will die or they will be merged. The latter would make sense: both products are columnar and while ParAccel has done some clever stuff with the optimiser Actian has focused on vector processing: I see no reason why you shouldn’t combine the two.

Of course that poses the question of whether Actian will hoover up any more of the smaller companies in the warehousing marketplace that have clever technology? Infobright would be the obvious target: decent customer base, not directly competitive (tends to target ISVs: and Actian knows very well how to market in that space), columnar, and it has some very clever technology in its Knowledge Grid. We shall watch and see.