Why was it decided in late 2011 that Ingres was not the right brand, given the company’s latest focus?
Ingres had around 30,000 customers around the world – a fantastic customer base, doing great things but in a traditionally commoditised market of transactional databases. We had already recognised that although that’s a terrific market it’s not a high growth market. We had been out and acquired the rights to a technology called Vectorwise, on the high performance analytics side.
That had come from watching lots of Ingres customers starting to do more and more and more intensive reporting, and figuring out that OLTP databases whilst great, are not necessarily optimised, and you have to throw a lot of hardware to deliver higher performance reporting, so we went and acquired Vectorwise.
As we engaged more with the Vectorwise customers we started to realise that even before big data was really on the market you could see that the rate at which newer customers were consuming data meant that a lot of questions were cropping up. They were doing more than just reporting on it – they wanted to get low latency notification of when things happen and want to be able to kick off different things. It was about the explosion in the consumption of different types of data. That opened a whole series of questions at the board level for us. We said OK, the real market that is exploding is this nascent big data.
Ingres is a commercially supported open source database. Do you want Actian to be considered an open source company?
No. I think open source in certain areas is very good. I love the business model. In one of the platforms that we’re announcing, the ParAccel Big Data Analytics Platform, a big part of our analytics capability is where we partner with KNIME, which is a powerful open source analytics suite. They are growing at a phenomenal rate. So although we are less and less open source as a vendor, I’m a very strong believer that genuine open source communities are very powerful. But it’s not core to us in terms of our own product development.
Do you see yourselves being at the data integration layer really, or do you also want to be an analytics player?
There’s a little bit of overlap. I picture it not as your traditional stacks from top to bottom but as a flow of data. Data is always flowing. Starting on the left with all the different data, whether it be transaction data from your databases, or your SAP system or your CRM system, all the way to device-based data coming from mobile data or wherever.
What organisations want to do now is capture that at a rate that is incomparable to the past. You used to have to go to IT and say I want to tap into this new source, load it into a database and they would design a schema. They don’t want to do that any more. All the customers that we see that I would call new age data companies, gaming companies, online ecommerce companies – they’ve got an entirely different view to it.
They essentially say they don’t care about the structure of the data: they want to consume it, which means connect to all the different sources; throw it in Hadoop, and then apply ETL [extract, transform and load] type processes. So I’m going to join data, I’m going to look for dirty data, convert data, before I finally offer it up. That could be in a database, in Vectorwise, in ParAccel or in Hadoop. For other things it doesn’t matter if it takes a day to get an answer. It’s the insight that’s powerful.
You’ve made several acquisitions in support of the new strategy: ParAccel, Pervasive, Versant…
Yes we realised that you need to be able to connect with data in the cloud, and once you’ve connected with it you need a massively scalable architecture to be able to process that data and present it up in the right repository for people to put their BI tools and analytic applications on top of it. That’s how we’re launching the Actian DataCloud Platform and the ParAccel Big Data Analytics Platform.
Versant is an object-based database. I remember when people thought object databases might one day replace relational databases but it never happened.
Object databases are seeing something of a resurgence. There’s a different use case. If I look at Ericson for example and how they use the Versant object database, you only have to look at the complex data models that they have. Everything to do with a mobile billing environment. No relational database would run with any sort of acceptable performance. Object databases are extremely good at representing certain types of data relationship.
I met the CEO of Informatica recently and his and your views on data integration for big data are quite similar. Do you see them as competition?
Informatica is by far the market leader in ETL capability, integration. It’s pretty high-end enterprise solution with associated pricing that goes with that. But it’s adopted by a lot of the market so any one of our customers may turn round to us and say they want to use Informatica to bring their data into our optimised data stores. That’s no problem, we have certification of tight integration with Informatica. The area where there’s overlap is our DataCloud because Informatica is trying to take their solution and put it into the cloud. Their challenge is that people expect a different economic model in the cloud so they need to be careful they don’t cannibalise their own products.
If I wanted to do big data integration wouldn’t I just turn to one of the bigger players like an IBM, Informatica or Oracle?
The core competitive advantage we have is a fully functioning, extremely high performance platform that is a price point that has opened up a huge potential market. Look at most digital media companies – they’re relatively small and yet they are processing more data than you and I could have dreamed of a few years ago. They’re using things like Hadoop, they’re looking at new types of technology. They’re never going to go to an IBM or an Oracle. They can’t afford a million-dollar price tag just to get started.
You made several acquisitions, is that the end of acquisitions for a while?
Whilst I catch my breath, yes. Acquiring companies is one thing. Successfully integrating them is another.
This article is from the CBROnline archive: some formatting and images may not be present.