Follow Datanami:
September 25, 2013

HP Says Big HAVEn Push Is Working

Alex Woodie

Hewlett-Packard was a little late to the big data party, which it entered in June with HAVEn, a collection of data analysis products it sells to customers directly and through system integrators and resellers. This week the company attempted to demonstrate that it’s caught up, as evidenced by the 120 partners it’s signed on to build and deliver HAVEn applications, as well as the availability of a new cloud-based version of HAVEn.

If you go to the main HP website, you won’t find a button that lets you download a trial version of HAVEn. You won’t see performance benchmarks for HAVEn versus other Hadoop or NoSQL database vendors, or case studies proclaiming how HAVEn has solved a company’s big data needs. And while HP does offer a single brochure for HAVEn (you can see it here), the product information for HAVEn is primarily spread out across multiple websites.

This decentralization is the direct result of HAVEn’s nature. You see, HAVEn isn’t so much a product as an integration concept. Look closely at the acronym, which stands for Hadoop, Autonomy, Vertica, and Enterprise security. (The lowercase n, by the way, stands for “n,” as in “any number of applications.” So much for specificity.)

But what HAVEn may lack in clarity and charm, HP hopes to make up with flexibility, industrial breadth, and technical depth–not to mention a whole bunch of development and integration services. Hadoop is an integral component of HAVEn, and to that end, HP works with the three leading Hadoop distributors, including Cloudera, Hortonworks, and MapR Technologies.

Depending on what industry a customer is in and what they’re trying to do with big data, HP will sell them one (or more) of three core engines that make up the HAVEn, including the Autonomy IDOL engine, the Vertica column-oriented database engine, and the ArcSight security information and event management (SIEM) engine.

In addition to having separate websites, Autonomy, Vertica, and ArcSight are three well-respected products that tackle different workloads. HP may have taken an $8.8 billion write off as the result of its 2011 acquisition of Autonomy, but there is consensus that its Intelligent Data Operating Layer (IDOL) engine is good at finding patterns within unstructured data, including text, voice, and video. Vertica, by contrast, excels at storing and analyzing semi-structured and structured data within a traditional SQL construct, while the ArcSight SIEM engine provides security-focused analysis of log data. HP is surrounds these engines with interfaces, making HAVEn both a technology wrapper as well as a marketing umbrella.

The HAVEn umbrella includes many components.

The flexibility of this approach means that HP doesn’t insist on storing your data within the various HAVEn engines. Instead it employs a distributed and de-centralized approach to data storage and integration, according to John Knightly, vice president of market development for HP Software.

“We have 700-plus connectors to enterprise data and information, not just to traditional database and applications and data warehouses, but to various machine logging sources as well as things like Sharepoint and Outlook, and all the places you might be storing unstructured human information,” Knightly says. “You can decide to put the information into Hadoop to analyze it, or you can leave it where it resides, index it, and bring it in for analysis.”

HAVEn isn’t a replacement for Hadoop, but can help extend Hadoop, he says. “It’s about embracing and extending Hadoop and making it easier to get more value out of it,” Knightly says. “There are use cases where you have batch analytics that maybe are find for Hadoop, but other [use cases] that require speed or more analytic processing around the meaning of human expression. Those are things our engines can do that might be hard to achieve in Hadoop.”

HP is one of the biggest enterprise IT companies in the world, serving the top corporations in almost every industry. HAVEn reflects that infrastructure approach, and because of that, it probably won’t be the right big data tool for a small company looking for a shrink-wrapped analytics platform–unless that company is ready to pony up some money to develop their own apps, or pay a partner for custom development. (HP says that some of its partners’ HAVEn solutions are aimed at midsize companies, in addition to large companies.)

“We see [HAVEn] as a platform on which you can build your own applications,” Knightly says. “People are building application-oriented offerings on it. Each integrator has its own methodologies, reference architecture, and best practices.”

The fact that HP doesn’t have a strong history of database development is a blessing, not a curse, Knightly says, because it allows HP to “bring new things the table. Customers see that we don’t have database or application agenda,” he says. “We can work with whatever they have in their environment.”

It appears this open-ended approach is resonating, to some degree, with HP’s partners. This week, Deloitte Consulting announced it’s using HAVEn to build a fraud detection solution. Other HP partners announcing support for HAVEn include Accenture Analytics, NEC Asia Pacific, and TATA Consultancy Services.

The company also announced a new cloud offering based on HAVEn. Instead of buying X64 servers on which to host Hadoop, the HAVEn engines, and the 700-plus connectors–which HP could certainly provide you with–customers can now simply pay HP to run and manage the HAVEn software for them.

HP says it’s delivering several HAVEn solutions in the areas of customer intelligence, supply chain and operations, and sensor data analytics. The solutions will be targeted at organizations in the media and entertainment, consumer processed goods, retail, travel and transportation, and public sector industries.

Related Items:

Getting to Optimal: An HP Vertica Query Tale

MapR Gooses HBase Performance in Pursuit of Lightweight OLTP

Is MarkLogic the Adult at the NoSQL Party?