Analysis: What Oracle-Sun Means for Business Intelligence and Data Warehouse
The new Oracle-Sun combination has compelling, and potentially breathtaking, business intelligence and data warehousing ramifications.
- By Stephen Swoyer
- February 24, 2010
With its acquisition of the former Sun Microsystems Inc. a done deal, Oracle Corp. recently outlined its vision. It's epic in scope; as an event, Oracle-Sun's presentation was epic in duration: officials took a full five hours to complete their pitch.
Although much of that was given over to a general appreciation of just how and why Oracle is good for Sun (or vice versa), the new Oracle-Sun combination has compelling, and potentially breathtaking, business intelligence (BI) and data warehousing (DW) ramifications -- starting, of course, with Oracle's ambitious Exadata Version 2 (V2) database system.
One Stack to Rule Them All
For starters, the post-Sun Oracle clearly positions itself as a single-stack proposition. That in itself isn't new, of course: in the DW space, for example, Teradata Corp. has been a single-stack provider -- comprising server hardware, database software, data warehouse special sauce, and an array of associated or complementary tools -- for about two decades.
In the general computing space, IBM Corp. can also claim to be a single-stack provider: it markets several Big Blue-centric database or data warehouse configurations (including pre-configured versions of its System x and System p systems running versions of the DB2 database optimized for either OLTP or analytic -- i.e., data warehousing -- workloads), and pitches single-stack platforms in its System i midrange systems (which come bundled with a System i-specific version of DB2) and its System z mainframes (which support a low-cost way to host both DB2 and data processing workloads).
It is IBM and System z -- and not a DW-focused player such as Teradata -- that is perhaps the best analog in this regard.
"The … surprise today was the strength of the all-Oracle stack message. Oracle CEO Larry Ellison … made it clear that the IBM of the 1960s was their integration model," wrote Gordon Haff, a principal IT advisor with consultancy Illuminata, of Oracle's event. "[P]romoting the benefits of buying a complete hardware and software stack designed to work together was one of the other overriding themes of the day."
The IBM of the 1960s was, of course, a mainframe-centric player. Even today, System z remains IBM's most specialized platform. It boasts features Big Blue simply doesn't support (or can't yet deliver) in other contexts. That's what Oracle plans to do with Sun: identify (or amplify) synergy or complementary areas and exploit them to differentiate Oracle-running-on-Sun from Oracle-running-on-any-other-platform.
This doesn't simply mean promoting a lower-cost combination of Oracle software and Sun hardware. It also involves playing a trump card: not only can and will Oracle-Sun function as a one-stop shop for all of an enterprise's computing needs, but -- from this point on -- many of Oracle's best, brightest, most interesting, or most differentiated technologies will be available exclusively in Oracle-Sun configurations.
Nor does this mean that Oracle (like Big Blue, with its POWER architecture) plans to extensively leverage Sun's UltraSPARC chip assets. That might seem obvious, but in the context of lingering doubts about Oracle's UltraSPARC strategy, it's far from a done deal. In fact, Oracle and Sun can make a compelling case for innovating on top of (and adding value to) standard Intel kit (see http://esj.com/articles/2009/04/28/sun-servers.aspx).
Oracle-Sun and Exadata
It's the DW arena in which the Oracle-Sun single-stack pitch could first be put to the test.
The test case is Exadata. It might seem naïve to accord such importance to what is -- or was -- a data warehouse system, but Oracle recently took Exadata in a very different direction. Although the revamped Exadata still has a distinct (and compelling) BI and DW pitch, it's now slotted for a greatly expanded role, too.
To recap, the first Exadata product -- Exadata V1, or what used to be called the Oracle Database Machine -- ran on a combination of software from Oracle and hardware from Hewlett-Packard Co. (HP). Exadata V1's differentiation relative to other analytic database systems came by way of its storage tier: its topology somewhat resembled that of a Netezza in that its storage tier (a cluster of what used to be called "Exadata" nodes) actually run queries (as opposed to simply passing raw disk blocks back to a cluster of Oracle database servers) -- see http://tdwi.org/articles/2008/10/01/analysis-strong-reactions-to-oracles-database-machine.aspx.
Exadata's -- and Oracle's -- differentiation now likewise centers on the storage tier. In Exadata V1, that storage tier was powered by HP; in Exadata V2 -- which Oracle announced just four months ago -- that storage tier is powered by Sun. Unless Oracle undergoes a drastic about-face, the Exadata storage tier will remain Sun-powered in perpetuity.
On the other hand, Exadata isn't yet (and isn't poised to become) an UltraSPARC-powered proposition. Instead, it runs the same 64-bit Intel Nehalem chips that Sun uses in its standard Open Network servers. Last year, Sun made solid state disk (SSD) and flash cache technology important components of its Nehalem-based Open Network Systems launch. Exadata -- with its 5 terabytes of Flash storage -- takes Flash to a higher level. According to analytic database specialist Curt Monash, Flash is what distinguishes Exadata v2. "Otherwise," writes Monash, a principal with Monash Research, on his DBMS2 blog, it's "just an annual refresh of Exadata Version 1 to include updated [Intel or OEM] components."
Nevertheless, Monash adds, Exadata is what Oracle is banking on to differentiate it in an increasingly commoditized analytic database market "Oracle doesn't make its storage-tier software available to run on anything than Oracle-designed boxes. At the moment, that means Exadata Versions 1 and 2. Since Exadata is by far Oracle's best DBMS offering -- at least in theory -- that means Oracle's best database offering only runs on specific Oracle-sold hardware platforms."
Exadata is just one of the pieces that Oracle hopes will underscore the strength of the new Oracle-Sun world. It's shaping up to be a big -- perhaps even a crucial -- piece, however. Industry watchers have already suggested that, post-Sun, Oracle could make a compelling pitch to enterprise CXOs -- particularly on the strength of its Exadata database systems because Exadata v2 -- the Sun-based Exadata variant -- is designed to support both OLTP and analytic workloads.
It's a not-so-complicated formula, according to IDC analysts Jean Bozmann and Matthew Eastwood, who note that Oracle is one of the most trusted -- and entrenched -- names in database and enterprise software. Sun is likewise one of the most trusted -- and entrenched -- names in enterprise hardware, particularly as a host platform for Oracle database systems. What's more, Exadata v2 has much broader appeal than its predecessor: Oracle positions it as a strong consolidation platform for all of a shop's Oracle database instances, OLTP- or data warehousing-oriented alike (see http://esj.com/Articles/2009/12/02/Exadata-Attracts-CXOs.aspx?Page=2).
That's one reason why Exadata v2 isn't a product offering in search of a market as much as it is a value proposition designed to address an existing (and pressing, say Bozman and Eastwood) market need.
"In effect, Oracle is making a strong statement about data transformation that could move at least some of the largest Oracle data warehouses in the world, now running on older SMP systems, into a new computing paradigm," they write. "This is significant because large Oracle databases and Oracle data warehouses are often one key reason that customers prefer to extend their current server hardware, lengthening server lifecycles for scalable servers, especially Unix servers."
The Single-Stack, Mixed-Workload Future
Oracle itself believes that this will happen. Monash, for example, cites an interview with Juan Loaiza, senior vice president of systems technologies with Oracle, in which Loaiza suggests that the "bulk" of Oracle's database business will transition to Exadata (or, as Monash puts it, to an "Exadata-like") technology at some point in the next decade. This isn't necessarily wishful thinking on Oracle's part, either. "Numbers-wise, this seems to be based more on Exadata being a platform for consolidating an enterprise's many Oracle databases than it is on Exadata running a few Especially Big Honking Database management tasks," Monash notes.
Other BI and DW watchers are more skeptical. "People are upgrading to Exadata on DW workloads, from what I've seen," says a prominent industry analyst who spoke on condition of anonymity. "I think this is a story that makes Oracle customers happy to stay on Oracle and pay lots more money to Oracle. Performance- and cost-wise, I'd never advise someone to go that route without considering other options, but there's something comforting about driving the same truck your father drove."
Author, veteran data warehouse architect, and consult Mark Madsen, a principal with Third Nature, has a mixed take on both Oracle's more ambitious Exadata pitch and the merits -- or potential attractiveness -- of the new Oracle-Sun Axis.
"I think that Oracle … may be right about the bulk of customers moving to Exadata simply because Oracle is flogging it as the path, and there are lots of companies with smaller workloads where it's feasible," he comments.
Madsen isn't quite persuaded by Oracle's mixed-workload message -- even (or especially) if it's yoked to an all-in-one Oracle-Sun technology stack. "Marrying hardware to software and compressing price is a potential with Oracle and Sun," Madsen concedes, stressing that there's a danger here, too: Oracle-Sun could go the way of IBM-DB2, which he notes is "rarely found running on anything other than IBM hardware and Unix … [or] Linux."
To the extent that both Gartner Inc. and IDC have IBM as the runaway Unix market leader, this isn't necessarily a bad bet, however.
More important, Madsen concludes, single-stack-mixed-workload isn't the direction in which the DW and broader database markets seem to be headed. "Running all the OLTP and BI workload on the same box is iffy unless they can dynamically partition and isolate workloads to keep one part of the system from stepping on the other. There's still a good reason for separating analytic workloads and data from the operational workload and data," he argues. "I don't see it going away. If [venture capital] investment in analytic databases is any sign, the future is separate data platforms for the two workloads."