Analysis: A Closer Look at HP's Acquisition of Vertica
HP seems ready to make yet another run at BI superstardom.
Over time, there've been more than a few head-scratching moves in the analytic database (ADBMS) space, such as Microsoft Corp.'s 2008 acquisition of the former DATAllegro Corp.
At first glance, the acquisition of Vertica Inc. by Hewlett-Packard Co. (HP) seems to invite second-guessing. After all, HP already partners with Microsoft Corp. in both its SQL Server Fast Track Data Warehouse and SQL Server Parallel Data Warehouse (PDW) programs.
In fact, HP took center stage just last month to tout its SQL Server bona-fides, announcing a new line of SQL Server-based appliance systems, including a new SQL Server-based offering for SME customers.
Moreover, HP already had an ADBMS offering of its own -- the former Neoview, a pet project of departed CEO Mark Hurd -- which it discontinued late last year.
HP was careful to position Neoview as an enterprise data warehouse (EDW), and officials invariably downplayed comparisons with ADBMS pure-play vendors (such as Vertica), instead designating Teradata as HP's and Neoview's competitor apparent. To some, Vertica reduplicates Neoview's product category (i.e., an ADBMS) and market niche (i.e., as a platform for high-performance analytics) -- just months after HP discontinued sales of that product.
Interviewed at TDWI's recent Winter World Conference in Las Vegas, HP officials wouldn't say much about the then-new Vertica deal, citing regulatory concerns. Giuliano Di Vitantonio, who heads global marketing for BI with HP, did discuss HP's decision to discontinue Neoview, however.
"That was a decision based purely on the commercial uptake on the technology. We didn't feel that commercially it was very viable," he said, adding that HP is "totally committed to supporting customers on the existing technology."
What about the possibility that HP's acquisition of Vertica could complicate its BI partnership with Microsoft? Di Vitantonio doesn't think so.
HP's strategy is to team up with strategic partners such as Microsoft and SAP AG -- across all technology domains -- and with both these partners and a handful of others (such as Informatica Corp. and MicroStrategy Corp.) in the BI and DW spaces. In BI and DW, Di Vitantonio pointed out, HP's primary emphasis is on the hardware and services side. "We come to this market [BI] with a very limited technology portfolio. At one time it was Neoview, now … it's something else."
Industry watchers seem divided about HP's Vertica gambit.
Jill Dyché, president of Baseline Consulting and a veteran -- stemming back to the 1980's -- of Teradata, thinks it a logical move, although she stops short of calling it a coup.
"HP buying Vertica isn't really that different from IBM buying Netezza," she says. "Vertica gives HP a lot of new opportunities in terms of licensing. It gives them a small yet established client base, and it gives them a whip-fast technology -- the buzz is that Vertica outperformed the other appliance players hands down in several recent benchmarks."
It also gives HP brand cachet. The Neoview product, Dyché points out, was based on "old" technology -- in this case, technology assets that HP inherited (by way of its acquisition of Compaq Computer Corp.) from the former Tandem Computer Corp. Vertica, on the other hand, has made its notional "newness" -- relative not only to "old" technologies such as Neoview but also to upstart ADBMS competitors such as Netezza and Dataupia Inc. -- a key part of its messaging.
Dyché thinks HP isn't done shaking things up in the BI and DW spaces.
"I would guess … [that] Vertica is one of several acquisitions, most of which are still in the offing, to enhance the evolving HP stack."
Veteran DW architect Mark Madsen, a principal with consultancy Third Nature, likewise expects HP to continue making BI and DW moves.
"HP can't afford to ignore the BI market. They hired several people knowledgeable about BI from SAP, so it was obvious that they were going to try something, but unclear what," he notes.
The acquisition also highlights the ongoing factionalization of the ADBMS space: three years ago, the ADBMS players that didn't market turnkey appliances tended to tout their willingness to partner with any of the big hardware OEMs -- including (in the cases of ParAccel Inc. and the former Greenplum Software Inc.) Sun Microsystems Inc.
As of 2011, however, most analytic database vendors have settled on one or two (at most) primary hardware partners. This was the case with Vertica, which Madsen suggests sold most of its ADBMS software on HP hardware. "I believe that most Vertica sales were either on HP or brought in [i.e., cultivated] by HP in some form," Madsen notes. "Dell has been working with Aster, Vertica had an HP relationship; almost everyone else has a hardware partner or was acquired."
In this respect, he notes, HP's acquisition of Vertica "seems one of the logical choices." It might also seem to smack of IBM Corp.'s acquisition of the former Netezza Inc. After all, Netezza based its TwinFin appliances on System x hardware from IBM, so few folks seem surprised when IBM ponied up $1.7 billion for Netezza late last year. Although there is a facile similarity here, Madsen contends, the two acquisitions are decidedly different.
"The difference between IBM-Netezza and HP-Vertica is that Netezza was bought by the [IBM] Software Group, and they seem to understand how businesses use information," he argues. Citing the Neoview failure, Madsen says he doesn't "see that same level of understanding at HP."
Madsen doesn't think an HP-branded Vertica ADBMS is necessarily inimical to HP's BI and DW partnership with Microsoft.
"If you look at Microsoft PDW, it's aimed [at] people who outgrow SQL Server and want to stay with it," he observes. "Vertica is a good move in that HP can control the software direction and it solves a specific need. It's unlikely companies will move to PDW if they aren't already Microsoft shops."
Kim Stanick, a principal with technology marketing consultancy Tekstur, thinks Vertica is a more focused (or less diffuse) product than Neoview.
Vertica founder Michael Stonebraker's vision "was to make data marts really simple, really easy to deploy and manage," says Stanick, who helped bring ADBMS upstarts DATAllegro and ParAccel to market.
"From what I understand, they do that -- they do that really, really well -- but this is very different from something like a Teradata, where you have lots of mixed workloads, lots of different kinds of queries," she points out.
HP always positioned Neoview as a Teradata killer, but Neoview lacked both the cachet and the credibility -- to say nothing of the critical mass -- to take on Teradata on its own turf. More to the point, says Stanick, neither Vertica nor Teradata nor SQL Server PDW is a one-size-fits-all proposition.
Prospective customers have many and varied needs. Ideally, she says, they'd select the DBMS technology that's best suited to these needs. That's why Stanick doesn't see HP's strategy with Vertica as conflicting with its existing BI or DW partnerships -- especially if it decides to position that product as an application- or domain-specific ADBMS-par-excellence (rather than, ala Neoview, as a Teradata-killer EDW platform).
"All of these different workloads are looking for the best price/performance, and one technology can't do all of them with the best price/performance [for each workload characteristic]," she notes.
Not everyone is so sanguine about HP's move.
One BI market watcher, who spoke on condition of anonymity, cited a conversation he had with a prominent BI industry veteran. "[Their] insight was that HP didn't understand why Neoview failed. They said this indicates that HP saw the failure not as a marketing, sales, consulting, or company problem, but as a product problem. It makes their approach obvious: replace the product," this person told BI This Week.
Veteran market watcher, consultant, and ADBMS expert Jos van Dongen, a principal with Netherlands-based Tholis Consulting, has a similar take. If HP believes Vertica vouchsafes it a place among BI and DW superstars, van Dongen suggests, it's only going to reprise the mistakes it made with Neoview.
"[Just] a database, columnar or not, will not give them any traction in the BI market, since HP is not seen as a player in that field," he says, noting that HP employed just such a tactic with Neoview, which "wasn't a bad product."
Van Dongen doesn't see HP's ADBMS bid as similar to those of Oracle -- which markets Exadata as its ADBMS centerpiece -- or IBM, which markets ADBMS offerings in the form of its Netezza and Smart Analytics systems.
"For both Oracle and IBM the situation was completely different: they both had a very strong BI stack already and now they've completed it with a very fast analytical database," he comments, adding that HP's acquisition of Vertica "only makes sense if more acquisitions in the BI space are announced shortly."
The deal also jeopardizes HP's relationships with other ADBMS players, van Dongen notes. It's true that ADBMS vendors tend to partner with one or two primary hardware manufacturers, but HP -- as the largest x64 hardware OEM in the world -- is primary partner to just about everyone.
"Kognitio, Paraccel, and others have a strong HP partnership. If I were one of Vertica's, competitors, I'd try to spread out more … like Microsoft does with [its] 11 hardware partners," van Dongen concludes.