HP's Half Step Back Into Data Warehousing

By Doug Henschen,  InformationWeek
Feb. 16, 2011
URL: http://www.informationweek.com/story/showArticle.jhtml?articleID=229218795

There's plenty of news in the data warehousing market this week, but the lead story is clearly Hewlett Packard's planned acquisition of Vertica.

Vertica's rivals are going about their business at The Data Warehousing Institute (TDWI) annual World Conference in Las Vegas this week -- SAP's Sybase unit talking up its strong showings in the recent Gartner Magic Quadrant and Forrester Wave reports, Kognitio announcing an MDX connector to support OLAP analysis, and Infobright debuting a database upgrade and fast data-loading option. More on that later.

The HP-Vertica deal has broader implications for data warehousing practitioners and vendors alike, even though HP isn't being particularly open or detailed about its ambitions.

The terms of the Vertica acquisition, set to close in the second quarter, were not disclosed. The deal is no shock in that it's another in a recent string of consolidations in data warehousing, with Sybase (and its Sybase IQ data warehousing product) acquired by SAP, Greenplum acquired by EMC and Netezza acquired by IBM all within the last year. But the deal raises plenty of questions about HP's big-picture strategy, implications for partners, and next acquisition candidates.

The key executive HP quoted in its press release on the Vertica deal is Shane Robison, executive vice president and chief strategy and technology officer. But he's not talking. Requests for an interview were denied and questions met with e-mailed responses by an HP spokesperson.

Vertica's Strengths

Among about a dozen independent vendors remaining in data warehousing, Vertica has a lot to offer HP. The company's column-store database delivers high data compression for efficient storage and fast querying in analytic applications. It's also low maintenance, able to get up and running quickly and maintain performance with less tuning than required by conventional relational databases, such as those from Oracle, IBM and Microsoft.

The database architecture also supports massively parallel processing on industry standard hardware, such as that offered by HP, so it can scale out to handle hundreds-of-terabyte, big-data deployments. Scalability has helped Vertical score high-end digital marketing and e-commerce customers such as AOL , Twitter, and Groupon.

8 Big Data Deployments In Detail
(click image for larger view)
Slideshow: 8 Big Data Deployments In Detail

Vertica has been innovator, introducing options including in-memory and flash-memory analysis for faster querying. It introduced Hadoop MapReduce connectivity to support customers interested in so-called NoSQL analysis of non-standard data and low-cost storage and data processing options. Vertical was also among the first to venture into cloud-based deployment options, and HP said this will support its Cloud Service Automation solution.

What's more, Vertica was already partnered with HP, and many of its more than 300 customers run on HP hardware. The database is compatible with HP servers and storage options and it will be a simple matter to configure optimized analytic appliances on HP hardware.

What Vertica is not, at least in real-world deployments thus far, is an enterprise data warehouse (EDW) capable of supporting thousands of users and mixed query loads simultaneously. Workload management features were recently added to the database to try to plug that gap, but the product is used almost entirely for analytic data marts. Neoview It's Not

Ironically, the deal comes just a few weeks after HP quietly confirmed that it is no longer selling the Neoview platform. Neoview was very much an EDW product that was intended to compete against the likes of Teradata and high-end Oracle and IBM deployments. By all accounts the product died at the hands of corporate neglect, high prices and poor product differentiation.

HP's spokesperson practically dismissed Neoview, describing it a "a proprietary data warehouse solution." In contrast, "Vertica is a scalable and standards-based solution and will provide next-generation business intelligence," the spokeperson stated. " We will work with Neoview customers to determine the most appropriate migration path."

Vertica is absolutely scalable up to hundreds of terabytes, but according to Gartner's latest Magic Quadrant report, the vendor has few customers with more than 100 users. A Neoview customer (like HP itself) would be hard pressed to handle that product's mixed query load and "tens of thousands of users" capacity on Vertica.

From a product standpoint, HP effectively abandoned one end of the market and embraced another. The gain is that Vertica's opportunity is far more mainstream. More than 90 percent of data warehouses contain less than 10 terabytes of data, by most estimates, and it's a domain where Neoview never made sense. Vertica does make sense for these smaller deployments, but it can also scale out in massively parallel processing deployments.

Given HP's Neoview management track record, one would hope HP will retain Vertica's (Bellerica, Mass.) management team and gives it a degree of autonomy. EMC and IBM have done just that with their respective Greenplum and Netezza acquisitions, and by most accounts those units have not suffered a post-acquisition malaise.

Will HP take the same approach? "Too soon to say," responded HP's spokesperson responded by e-mail. "Our intention is to acquire an operational business, and we expect that we will need Vertica employees to continue their current functions to maintain this business"

The spokesperson added that HP expects Vertica to become part of HP Office of Strategy & Technology (OS&T), led by Shane Robison. That unit's focus is "to provide small business units the resources and support they need to focus and grow quickly." For example, OS&T is currently incubating HP Visual Collaboration and Cloud Services business units.

8 Big Data Deployments In Detail
(click image for larger view)
Slideshow: 8 Big Data Deployments In Detail

What About Partners?

The big complication in the tie-up with Vertica is the likely impact on HP's partnerships with Microsoft and SAP. On the first front, Microsoft and HP recently announced a series of joint "Converged Application Appliances" designed to speed and lower the cost of data warehousing, analytic and messaging deployments.

According to HP's spokesperson, the company sees partnerships as "complementary" to the HP-Vertica deal, but it's hard to see how, with the possible exception of the Exchange Server messaging appliance. It's not as clear cut as Microsoft products being for Microsoft shops. Microsoft clearly has ambitions, with its SQL Server Parallel Data Warehouse, to move beyond its current customer base. I've talked to plenty of customers who have considered both column-store and row-store databases for the same application. AOL and Provisio, for example, both considered column-store and row-store, and the latter replaced a Microsoft SQL Server deployment with ParAccel, a column-store competitor to Vertica.

HP's alliance with SAP would seem to be unshakable, but it can't be sitting well in Walldorf, Germany, that HP is buying the arch rival of Sybase's Sybase IQ product. Vertica has been going after Sybase customers and openly criticizing Sybase's IQ product for years.

One slam Vertica has used in recent years has been to site Sybase IQ's lack of support for massively parallel processing (MPP). But that recently changed when Sybase added an MPP option. Sybase, like Vertica, has also been working on supporting faster data loading, mixed query workloads and larger numbers of users. Sybase's progress toward handing EDW deployments was specifically cited by Gartner in its recent Magic Quadrant report on data warehousing vendors.

Infobright is another vendor working on the historic drawback of column-store databases: slow loading speeds. Infobright introduced a Distributed Load Processor option this week to speed loading to up to one terabyte per hour. Fast loading is essential to deliver near-real-time analysis of data, and it's a challenge Vertica addressed in 2009 with its FlexStore feature.

Kognitio's introduction of an MDX connector for OLAP will appeal particularly to shops using Microsoft Analysis Services for online analytical processing. The connector will effectively eliminate the need to build cubes as the Kognitio database itself will serve up multidimentional data for fast analysis.

Will Vertica respond? These are the kinds of choices HP will have to make that may put it in closer competition with Microsoft offerings.

Divided Loyalties

Where Microsoft and Sybase are concerned, the loyalties of HP customers, not to mention HP salespeople, are likely to be tested. Historically customers choose the database first and the hardware second, but appliances are blurring those lines. Make no mistake, there will be deals in which HP-Vertica and Microsoft and HP-Vertica and Sybase, and SAP company, will be competing for the same customers.

8 Big Data Deployments In Detail
(click image for larger view)
Slideshow: 8 Big Data Deployments In Detail

As for HP's broader ambitions, the spokesperson offered only a broad statement about "continued investment in enterprise information optimization," noting he couldn't get into specifics in a pre-earnings quiet period.

I wouldn't expect HP to divulge any detailed plans, but the acquisition of Vertica is only a half step back into data warehousing. To compete with the likes of IBM, Oracle and Teradata, HP will have to somehow address the mixed-query, high-user-number EDW market it gave up on -- if, indeed, it intends to compete at the top end of the market.

And beyond databases and appliances, HP must also address the gaping holes in data integration, data quality, business intelligence and analytics portfolios. For that, we'll have to wait for yet more acquisitions.

转载于:https://www.cnblogs.com/mengheyun/archive/2011/02/18/1957814.html

评论
添加红包

请填写红包祝福语或标题

红包个数最小为10个

红包金额最低5元

当前余额3.43前往充值 >
需支付:10.00
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值