Data Warehouse Players Get Business-Savvy
The data warehouse appliance -- that wunderkind of the data management field -- might finally be growing up.
- By Stephen Swoyer
- May 21, 2008
The data warehouse (DW) appliance -- that wunderkind of the data management (DM) field -- is growing up.
At last week's TDWI World Conference, held in Chicago, a bevy of vendors announced new DW appliance packages -- with a twist.
In the past, several vendors have promoted the DW appliance as a technology which effectively sells itself -- e.g., as a turnkey panacea for the performance, capacity, and cost issues that frequently combine to make high-end data warehousing so expensive.
At TDWI last week, however, three vendors (Dataupia Inc., ParAccel Corp., and Sybase Inc.) unveiled new DW appliance packages that they say are designed to address specific business problems.
Dataupia's offering bundles the company's Satori server appliance with best-of-breed data visualization software from Tableau Inc. ParAccel announced a new analytic appliance offering it co-developed with storage giant EMC Corp. Sybase unveiled an all-in-one, best-of-breed analytic appliance (complete with Sybase IQ, Sybase PowerDesigner, and Sybase ETL; BI software from MicroStrategy Inc.; and System p RISC/Unix servers from IBM Corp.). In all three cases, the vendors are emphasizing business practicality and business usability in addition to better and cheaper performance.
Dataupia: Out-of-the-Box Data Visualization
Of all of last week's partnership news, Dataupia's announcement was perhaps the most intriguing. Data visualization seems on the cusp of going supernova while the DW appliance itself is verging ever closer to mainstream acceptance.
Is an out-of-the-box data visualization offering running on top of a Big Data appliance such as Satori Server the kind of hybrid solution that could push both segments over the edge? Perhaps, says Kevin Brown, who heads up Tableau's new alliances and partnerships effort. Tableau partners with at least one other DW appliance player (Netezza) and is currently preparing a bundling arrangement with a prominent publicly-traded player in the high-end data warehousing segment. (That deal should close sometime this week, Brown indicates.)
Brown argues that such arrangements are mutually beneficial: the bigger, faster, cheaper DW appliance model plays into Tableau's strengths -- particularly when it comes to pulling in data from various sources (including non-traditional data types such as map overlays, geospatial data, and so on). Tableau's visualization technology -- which experts such as Stephen Few, a principal with Perceptual Edge, recognize as a cut above the basic data visualization capabilities marketed by Business Objects, Microsoft, and other vendors -- amounts to a killer app for the DW appliance.
"This whole trend toward big data being optimized for visualization, it's all toward our sweet spot. The more data [we can consume], the better. If you give [Tableau] fast access to lots of data -- in this case, to terabytes of data -- we really get a chance to show what we can do," he comments.
What does Tableau's partnership with Dataupia entail? Call it out-of-the-box data visualization, says Brown: "The performance of Tableau is completely guided by the performance of the database server itself. The results of queries come back in graphical views of data, but that [the visualization of the data] isn't the bottleneck. The bottleneck is that fast access to the data, so in the case of what we're doing with Dataupia, you'll be able to visually query in real time billions of rows of data just like you're looking at an Excel spreadsheet."
ParAccel Emphasizes Manageability, Availability, Disaster Recover-ability
ParAccel's analytic appliance venture with EMC is a slightly different beast. Seventeen months in the making (EMC first approached ParAccel about co-designing an analytic appliance last January, according to officials), ParAccel bills the Scalable Analytic Appliance as a business-safe take on the DW appliance. By "business-safe," ParAccel doesn't necessarily mean chock-full-of-application-software, either.
Instead, says VP of marketing Kim Stanick, the Scalable Analytic Appliance bundles her company's columnar database software with EMC's NaviSphere SAN management software, Clariion SAN disk arrays, and hardware from an independent OEM (in this case, Dell Computer Corp.).
The value-add for the business includes built-in fault-tolerance (FT), disaster recovery (DR), and SAN management capabilities (including reconfiguration and provisioning), thanks to EMC's best-of-breed NaviSphere software.
"This [Scalable Analytic Appliance] is a really cool convergence opportunity," Stanick argues. "It's marrying these best-of-breeds [EMC Clariion SAN storage and NaviSphere management software] with our best-of-breed [analytic warehouse], so you get best in class, high availability, and disaster recovery -- which are so important to business customers -- and [with] which, frankly, our competitors simply can't compete," she continues.
"It used to be if you got one of these [DW appliances], you had to do this [availability and disaster recovery] on your own." Most appliance vendors tout their parallelism as a DR capability unto itself, Stanick says, but NaviSphere gives ParAccel a sophisticated DR feature set, including geographically dispersed replication and recovery capabilities.
The EMC/ParAccel Scalable Analytic Appliance also raises a couple of interesting possibilities relative to EMC's own product offerings. Consider EMC's Documentum content management solutions, for example: with a canned analytics offering, Documentum customers will be able to bring plug-and-play analytics to bear against unstructured and semi-structured information housed in Documentum content management repositories.
"It can lead to a lot of really interesting use cases," Stanick suggests.
Sybase Takes the Appliance Plunge -- With a Twist
In many respects, Sybase's Analytic Appliance is the most ambitious -- and, arguably, the most credible -- of all of the appliance entries that debuted at TDWI last week. After all, it comprises Sybase's vaunted IQ columnar database; Sybase's PowerDesigner data modeling tool (one of a pair of industry-leading modeling tools, the other being ERwin from CA Inc.); its fledgling ETL tool (which Sybase acquired from the former Solonde); BI software from MicroStrategy; and System p RISC/Unix hardware from IBM.
Sybase isn't pushing a technology montage, asserts senior product manager Andrew Neugebauer: its Analytic Appliance is optimized for ad hoc query and analysis, comes with a built-in services component (via mLogica Inc., a tripartite Sybase-MicroStrategy-IBM services partner), and promises rapid time-to-implementation.
"What we're delivering is a complete, end-to-end analytic solution," says Neugebauer, who quickly concedes that the expression "complete, end-to-end" tends to make one's eyes glaze over. In this case, he asserts, it's an apposite characterization. "A lot of people say end-to-end, yes, but what we're offering [customers] here really is a complete solution."
"If you think about what you need to do to analytics today, you need a design tool to figure out all of the different databases you're going to work with and how they're structured. That's what PowerDesigner gives you. You also need to figure out a way to do your connectivity, your transformations -- that's Sybase ETL. And you also need your high-performance analytics, which is Sybase IQ," he continues. "Then you need a front-end tool to actually do the analysis, to get some output. That's MicroStrategy, and that's in the package, too. Then [mLogica] installs all of this and tunes it on top of IBM's Power servers."
The days of buying technology for technology's sake are over, Neugebauer maintains; customers don't so much want cool or cutting-edge stuff as solutions -- practical, business-ready solutions that work.
This is particularly true in the area of ad hoc query and analysis, where -- argues David Jacobson, senior director of field marketing for Sybase -- customers are still ill-served by technologies that have poor performance, don't scale, offer little in the way of usable front-end interactivity, or are expensive.
Out-of-the-box DW appliances are fast and cheap, Jacobson concedes, but because they lack BI or analytic front-end software, they don't comprise solutions unto themselves. "[T]he value to customers is closing the gap between the hardware [vendors] and the various software partners, and IBM has been an extraordinarily good partner to enable us to close those gaps," Jacobson observes. "If you can close the gaps between the design tools, the ETL, the analytics, the front-end [tools], and the hardware, and if you can do that in a way that reduces the risk for customers, you're going to add tremendous value."
There's also value in delivering a centralized management experience, he argues, and that's just what Sybase, MicroStrategy, and Big Blue have done. "We've come together to build a management console that overlays your entire [analytic appliance] infrastructure to do all of the administration necessary, so it makes management of all of this easier," Jacobson says.
The Circle Is Complete
Veteran data warehouse architect Mark Madsen is encouraged by what's happening in the DW appliance segment. One of the reasons Dataupia, ParAccel, and Sybase are pushing new all-in-one offerings is because the appliance market itself is changing: Netezza, DATAllegro, and GreenPlum have already plucked the low-hanging fruit (i.e., customers with big performance problems), so today's appliance buyers tend to have other needs.
"The appliance market has some interesting trends due to crowding, immature technology, [and] a new base of novice customers as opposed to people with big performance problems," he notes, citing a definite trend away from the appliance as a "database-in-a-box to more of a complete solution."
In this respect, Madsen argues, the circle is complete. Sort of.
"The verticalized data warehouse-in-a-box is coming back," he says. "It's weird that I worked on two generations of data-warehouse-in-a-box and [that] we're on the third now. I think the third generation is getting it right -- though I'm not sure whether an appliance or software makes much difference."