HP promises real-time decisions from business intel

HP has released additions to its year-old NeoView Busines Intelligence (BI) software stack. They include real-time uploading of  data and analytics to provide immediate operational BI.

John Miller, VP of HP's BI unit explained, "Typically BI is used for running affinity studies and large complex queries, we're talking about bringing information from lots of different sources together and analysing it in real-time.

"For instance, if a call centre agent can see that a caller with a propensity to churn has been checking call rates and plans online, and checking out their account status via their phone, then the operator will probably want to route them straight to someone who specialises in customer retention."

He added, "Put another way, it's about allowing the front line to deliver better service and value to the customer: its about getting information to the right people at the right time."

How is this possible when HP claims no-one else is offering it‾ Miller said, "We're not competing in areas like analytics, Our job is to provide a platform that is capable of massive parallel processing - share nothing.

"Some parallel processing engines have pooled or shared memory, we split up data so that autonomous nodes do the work and then share the information gleaned. This means it can be scaled up to include hundreds of processors. To get economies of scale, we use conventional mainstream IT processors. The software leverages them, optimally, around the database."

How right HP has got its approach remains to be seen, but this is an area where data as a service (DaaS) is gaining some traction from the likes off relative upstarts like Kognito, which, among other things, has BT in the UK as a customer where it collects competitive tariff information and updates BT's own charges based on them. The company claims it uploads large volumes of data within 14 mintues and carries out data transformations in six.

The key questions are: how much is real real-time information worth to an operator (a very difficult question to answer with any accuracy); how fast is HP anyway, how do DaaS versus NeoView compare in cost; is this something operators will be glad to offload instead of trying to run in-house‾

Miller won't reveal a name, but says that it has and is running an extensive trial with a large Asian mobile phone operator that data warehouses some 3.2 billion call detail records (CDRs) per hour - via HP's NeoVision software -every hour into massive databases. He stresses that in addition to this and running concurrently are fraud analytics and detection, and customer service handle some 24,000 queries per hour too.