By using website you agree to our use of cookies as described in our cookie policy. Learn More


Analytics Trends Play to Hewlett-Packard's Strengths

HP officials cite a number of advanced analytic trends that they claim play directly to their strengths.

When it comes to analytics, HP may be sitting pretty.

Officials from Hewlett-Packard Co. (HP) started aggressively pushing advanced analytics at a time -- summer, 2008 -- when few outside of the former Applix Inc., SAS Institute Inc., the former SPSS Inc., and Teradata Corp., and several analytic database vendors were out in front with it.

Although by autumn, advanced analytics had progressed, few BI professionals seemed to recognize the trend.

HP's John Santaferro was an exception. During a wide-ranging interview with BI This Week, he discussed the emergence of new and altogether more demanding kinds of analytic workloads.

At the time, Santaferro also talked about a transition to more demanding BI landscape, citing HP's Neoview DW appliance's place in that environment. "[W]hat you have happening more and more is you have these business analysts pounding the data warehouse trying to get their information out of it. They can do this because [products like] Neoview support these kinds of activities. So when you give them more [freedom], whether it's performance in the back-end or [flexibility] with their own [front-end] tools, they're going to use it."

He explained that "our customers are just pounding Neoview with these extremely sophisticated, transactional-type queries," and expected to see "a lot more of this [kind of activity]. Right now, we're in the midst of moving from a kind of old-world-style BI to a much more demanding, much more sophisticated new-world-style."

Santaferro was somewhat ahead of the curve. Microsoft hadn't yet purchased DATAllegro Corp., Oracle Corp. hadn't yet unveiled its Database Machine, and venerable data warehouse (DW) appliance pioneer Netezza Inc. hadn't yet gone the (mostly) commodity route.

To be sure, most of the analytic database players were in place, players such as QlikTech (which markets a desktop-based analytic studio, dubbed QlikView) had made noises on the advanced analytic front, and -- in September of that year -- desktop analytics specialist LyzaSoft would make a big splash, debuting its Lyza analytic studio (even as Microsoft Corp. was then prepping its column-store-on-the-desktop announcement, Project Gemini), but a lot of folks in the BI and DW industries were struggling to articulate just what was driving the need for bigger and brawnier DW systems, to say nothing of a richer, more self-serviceable, altogether more ad-hoc-able class of BI front-end tools.

Staying the (Advanced Analytics) Course

We're still in the process of transitioning from old-world-style to new-world-style BI, according to Santaferro and other HP officials.

The difference, almost two years later, is that just about everyone understands what's going on, even if a number of vendors have only the most oblique of potential (advanced) analytic applications.

HP, Santaferro and others say, has demonstrable analytic background: for example, it purchased data warehousing (DW) consultancy Knightsbridge Solutions in December of 2006, shortly after it unveiled its first attempt at a Teradata-like enterprise data warehouse (EDW) platform, Neoview.

That offering has itself evolved substantially over the years, but then, as now, officials position it as an advanced analytic platform, a la Teradata.

Although officials are still wary about talking on the record about Neoview's customer demographics or sizing, they're altogether less hesitant when it comes to talking up its prospects.

Simply put, says HP marketing manager Vickie Farrell, several advanced analytic trends -- such as a growing appetite for data mining and predictive analytic technologies, as well as an emerging preference for in-memory and in-database processing -- play directly to HP's and Neoview's strengths.

"The real trend now is doing it in the database, as opposed to pulling the data out to do it on a separate platform, so performance limitations are being somewhat overcome by in-database memory processing," she comments. "With in-memory, when Neoview processes queries, it does not pool the interim results to disk like some other products do. It just keeps it right in memory. Some of our competitors do that, too. What we're seeing [a demand for] is in-database [processing]."

Right now, Farrell concedes, HP isn't alone in this respect: Teradata likewise partners with SAS -- the two companies announced plans to embed SAS analytics inside of Teradata Warehouse almost 18 months ago -- and Netezza has talked about in-database analytics, too. HP (like Teradata) says it's open to partnerships beyond SAS, however.

"We've looked at a number of options, we're actively looking, but I don't think we have any public statement at this time about that, so I can't discuss anything in detail, except to say that we're looking [at other options]. We really believe that's [i.e., in-database] is going to be an important area for us," Farrell comments.

Concomitant with the growth of in-database analytic alternatives, there's been an attendant leveling of the advanced analytic playing field.

The upshot, Farrell argues, advanced analytics aren't solely the purview of players like SAS or SPSS any longer.

"A second problem [with respect to customer adoption of technologies like data mining, statistical analysis, or predictive analytics] was always how hard is it to do these things, and we are again seeing the emergence or increasing adoption of tools that are easier to use than SAS," she indicates.

There's also the problem of the consumption, digestion, or management of data -- particularly of the huge volumes of data (often in the double- or even triple-digit-terabyte range) that are associated with advanced analytic requirements -- which is a trend that Farrell says also plays to HP's strengths. "The third problem is the organization of the data, the level of quality, the ability to integrate data from multiple sources, the ability to accommodate all of these different kinds of users on one system. In the past, these were the kinds of things that limited what we could do, [analytics-wise]. That just isn't the case now," she argues.

For this reason, she doesn't discount the success of analytic database players like Netezza, Aster Data Systems, Dataupia Inc., Greenplum Software Inc., Kognitio, ParAccel Inc., Vertica Inc., and others. Such offerings comprise tactical responses to advanced analytic requirements, Farrell argues. Eventually, she contends, most large customers will opt to take a strategic approach to analytics; that's where HP has a role to play, she maintains.

It's a familiar argument. It's in fact a pitch that smacks of Teradata's own messaging: like Teradata, HP and Neoview trumpet an ability to support mixed workloads; address a constellation of related data management issues (including data integration, data quality, and master data management); and accommodate virtualization, cloud computing, data center modernization, or other cutting-edge initiatives. Farrell invites the comparison -- even as she claims that HP can outclass Teradata in these and other respects.

"We're seeing a proliferation of these data marts," she comments, meaning analytic database platforms from Netezza and other specialty players. "What it is really is this combination of analytic data marts and operational data stores: they [i.e., customers] don't necessarily call it an ODS, but that's in fact what it is." In this respect, Farrell invokes HP's manifold hardware, software, and services bona-fides, a combination that she claims even Teradata can't match. "Our position is that Neoview, because it has an OLTP database heritage, because it came from [the former Tandem Corp.'s] NonStop [heritage], because we've meanwhile added a brand-new optimizer that's really purpose-built for data warehousing, because of our [HP's] rich hardware heritage--; our position is, why not do it all in the same platform, because Neoview can support the sort of high-throughput transactional queries that an OLTP database can support, but at the same time, you can use [Neoview] to do the long-running strategic queries, too."

TDWI Membership

Get immediate access to training discounts, video library, BI Teams, Skills, Budget Report, and more

Individual, Student, and Team memberships available.