Oracle’s Exalytics: Old Wine in New Wineskins

On October 2, 2011, Larry Ellison announced the latest in his “Exa” parade, called Oracle Exalytics. Here is my take on the announcement and why it is nothing more than the proverbial “old wine in new wineskins” (actually, very expensive hardware wineskins):

First, some context. In September 2010, during the Oracle earnings call, it was stated that SAP’s move into building an in-memory database (SAP HANA) was SAP’s biggest mistake, and Oracle vowed it would beat SAP to market with an in-memory database appliance. SAP was first to market in November 2010, shipping SAP HANA, a state-of-the-art, in-memory analytic appliance, with innovative analytic capability to handle today’s “Big Data” problems in real-time. Now, in October 2011, Oracle has finally shipped Exalytics, which stakes claims to in-memory analytics —nearly a year after SAP launched SAP HANA. Besides being first to market, SAP HANA is also significantly more innovative, as I will describe.

  • The Exalytics architecture just squeezes two separate databases – TimesTen (for query optimization) and Essbase (for planning), along with Oracle BI server (OBIEE 11g) – into a hardware-intensive appliance (4-socket, 48-core server with 1TB of DRAM, locally attached disks, along with an Infiniband connection to Exadata). Mark Rittman outlines its architecture here. This is an out-dated architecture that is overly expensive for what it seeks to achieve. For those of you who are more technically inclined, here’s a bit more detail; and non-techies can skip ahead to my next bullet point it you wish. To understand why Exalytics is just “old wine in new wineskins”, you’ve got to look inside it. Exalytics contains TimesTen, built 15 years ago in 1996, and out of the picture since Oracle acquired it in 2005. Exalytics also contains Hyperion Essbase, another dying OLAP technology built 20 years ago. TimesTen has no real column-based capability. Essbase even with all its “optimization” cannot efficiently run in-memory – you still have to do pre-calculations and pre-aggregates, with no ability to do calculations on the fly. You’d have to limit how far the Essbase calculations propagate to ensure performance doesn’t blow up, and insert operations force the indexes in the database to be rebuilt, thus ruining performance (NOTE: even Cognos TM/1 does OLAP in-memory better than Essbase).
  •  One would have expected that the in-memory query-optimization and planning capability could all be designed right into Exadata, rather than requiring a completely separate new machine in the form of Exalytics. But my guess is that because TimesTen and Essbase can’t sit efficiently inside the Oracle database architecture data has to be moved from Exadata (for transactions) into Exalytics (for analytics). There’s nothing real-time about this “patchwork-quilt” architecture. SAP HANA was engineered from day one, as a purely in-memory analytic platform. SAP HANA achieves a much cleaner architecture for query optimization, planning, advanced business logic and predictive analytics, all in one software platform running in an in-memory appliance. Hasso Plattner has described this in his SAPPHIRE NOW lectures as “collapsing the layers,” and it is a significantly more innovative architecture. The future of the real-time enterprise collapses layers of OLTP, DW, OLAP into a unified architecture, rather than requiring two or three separate machines. And the performance differences are “TimesHundred,” not merely “TimesTen”! Clay Christensen, professor from Harvard Business School, called such innovation, “disruptive innovation,” because it disrupts the status quo. One analyst watching Larry’s keynote agreed on Twitter – @DanVesset, “Oracle’ success with Exalytics, Exadata will maintain status quo. SAP’s success with HANA would be more disruptive to competition.”
  • The future of such innovation is in software, not in hardware! While Oracle requires their customers to use exorbitant hardware, SAP’s focus with SAP HANA is customer-centric innovation in the software stack, combined with an open hardware partnership for SAP HANA with IBM, HP, Dell and others. The open partnership with these hardware partners brings greater choice for customers in the market. Customers do not want to be locked into one hardware vendor. It’s the hardware innovation by an ecosystem of players that brings lower TCO at the rapid rates of Moore’s Law. SAP HANA is a perfectly “engineered system,” even if the hardware is actually built by any of the above named hardware players. Hardware partners are incentivized now to partner with SAP, in many cases to replace Oracle database systems.
  • When Exalytics was introduced on Sunday, October 2, no customers were mentioned. This social media tweet by @MKrigsman poignantly stated, “Oracle did not bring real customer case studies — #SAP did. SAP also understands importance of linking tech to business”. On Monday, Oct 3, Oracle did bring up 4 beta customers of Exalytics, but interestingly those customers were eager to point out that Exalytics showed better performance than Exadata, thus either obviating the need for Exadata in the first place, or reinforcing the fact that you need every “Exa” sibling, in the quest for true performance! The point remains, every CIO should ask Oracle this: “How many such Exa-boxes are you going to keep making me buy? First Exadata, then Exalogic, now Exalytics—I thought this was about reducing TCO?

It is clear that Oracle, with its obsession to create its Exa fleet of hardware boxes, is missing what customers really want and need – innovation, not consolidation!

In stark contrast, SAP has intensified its focus on customer-centric innovation, in five essential categories that individually and collectively emphasize our belief that “innovation for the future” is better than “consolidation of the past”:

1) Business applications – with leading industry flavors

2) Business analytics – with leading industry flavors

3) SAP HANA – both platform and applications, with performances of 100X and more, not just 10X

4) Mobility – showing the power of speed and innovation for the next generation

5) Business cloud – with a family of modern applications and platform, all on demand

CIOs, the choice is yours: sign up for Oracle’s Exa parade and its never-ending set of hardware purchases, licenses and maintenance, and it will increasingly lock-in control over you and your future, with antiquated architectures. SAP’s focus on innovation, with solutions like SAP HANA, feature a simplified architecture, fewer layers, more choice of hardware partners and vastly greater performance, all at significantly lower cost of ownership.

SAP Executive Board Member Vishal Sikka has also posted a video-blog on this, that I’d encourage you to read with more detail and his perspective here.