Ever since Leo Apotheker assumed the reins at Hewlett-Packard (
HPQ) – assisted by Ray Lane on the board – we have been awaiting the first hint that the new regime would take HPQ up the enterprise software food chain.
Apotheker has finally dropped the other shoe: HPQ has signed a definitive agreement to acquire Vertica, one of a growing family of columnar analytic databases. The announcement came barely two weeks after HPQ finally cut Neoview loose. It’s a pretty logical course, as Neoview technology was dated; inherited from HPQ’s acquisition of Compaq, which in turn swallowed Tandem; you get the picture – Neoview was the odd man out. Out with the old wave, in with the new, as Vertica is one of a group of emerging analytic database vendors. in this case, it is one of the first, but not the only, column-oriented, massively parallel analytic database player (MPP) out there.
Although it’s Valentine’s Day, we’ll avoid the temptation to call this a sweetheart deal. But it is a shrewd acquisition for HPQ because, as one of a group of emerging players, it is a good modest-sized first step for Apotheker’s software strategy. Vertica isn’t so big to that the acquisition cost would break the bank, nor a large enough player to add major organizational disruption. The move also has merit for other reasons: first, as a consumer of commodity hardware, Vertica provides a good fit for HPQ to add value with more appliances. And as implicated above, it provides HPQ a foothold into the direction that analytics are going: towards big data (the ability to process Internet-scale troves of data) and Fast Data (the ability to process huge chunks of data in real time). Vertica promotes its ability to scale and provide near real-time performance.
One of many companies founded by Ingres and Postgres pioneer Michael Stonebraker in 2005, Vertica has amassed roughly $30 million in venture backing and even in recession-wracked 2009, found its business tripling; the company now has over 150 customers. So the company is clearly a going proposition.
Of course, Vertica is hardly unique here. For instance, it has a handful of rivals that are also commercializing vertical database technologies, with one, ParAccel, having drawn almost as much venture backing. Column-oriented technology has proven useful for analytic applications because (1) the need is not for retrieving individual records (rows) and (2) scanning by columns rather than rows saves lots of process overhead. Admittedly, the technology is hardly new; Sybase (SY) acquired a startup in the mid 1990s, with Sybase IQ being an established product with a large installed base. But it is based on older SMP technology, as MPP was not economical back in the 90s. MPP-oriented columnar databases emerged over the past five years thanks to cheapening commodity hardware and increasing memory density — bringing massively parallel implementation within the budgets of mere mortals. Vertica also rides a wave to the handling of big data, as it (like others) provide support for Hadoop support for access to humongous Internet-scale data sourcing.
HPQ had to make a move like this, and not because they now have an enterprise software guy in charge. With IBM (NYSE:IBM) and Netezza, Oracle (NYSE:ORCL) and Exadata, and EMC (NYSE:EMC) and Greenplum, HPQ’s rivals were rapidly buying up analytic database appliance providers (we’re waiting for Dell to make its move). They were not buying up (or developing) analytic database hardware systems just because they were there, but because there is an increasing feeding frenzy for the ability to process big data and fast data as a result of:
- Commodity hardware and lower cost memory, as mentioned above.
- Actual use cases that bring tangible benefits, such as sentiment analysis of social media.
- Use of predictive analytics becoming, not a competitive differentiator, but a cost of doing business.
- Open source technologies and APIs from Internet giants like Google, Facebook, and yahoo bringing big data to the masses.
This is not the first time that BI has entered a commoditization wave; it’s just that the bar keeps rising. Two examples: Microsoft’s packaging of OLAP as an option to SQL Server commoditized multi-dimensional databases; and enterprise applications began packaging their own analytics. In neither case did the market for advanced analytics disappear, just that the SAS’s, Cognos’s, and Business Objects of the world had to advance their capabilities. To this, add support for inline hardware platforms that perform the heavy lifting of processing huge amounts of data, or processing data in (or almost) real time. History will repeat itself again.
For HPQ, this is a logical first step towards raising its profile in the enterprise software space. It still has significant heavy lifting ahead of it, as it must balance the ongoing converged infrastructure initiatives that are melding 3COM, 3PAR, and Palm into the business. HPQ’s danger is getting distracted. With Vertica, it is a modest, digestible step and proof of concept as to whether HPQ can aim higher in its enterprise sales pitch.
Disclaimer: Author is an analyst with Ovum; the opinion stated does not necessarily reflect that of Ovum.