Skip to Content

This piece was co-written with Jonathan Becher, CMO of SAP

On Sunday night, during Larry Ellison’s lackluster Open World keynote, Oracle finally announced their in-memory analytic appliance dubbed Exalytics, validating SAP’s strategy but failing to impress many observers.  Before I comment on the announcement, it’s worth a walk down memory lane to better understand what was behind their announcement.

In May 2010 at SAPPHIRE NOW, SAP first announced its technology vision of non-disruptive innovation by combining in-memory, cloud, and mobility. At the heart of this strategy is SAP HANA, an in-memory database.  SAP HANA started its life out as a real-time analytics attachment to SAP Business Suite, SAP NetWeaver Business Warehouse, and SAP BusinessObjects – delivering real-time analytics to SAP customers.   

At the time Larry Ellison poked fun at Hasso Plattner, an SAP co-founder, saying that Ellison wanted to know what pharmacy Hasso was getting his drugs from.  A few months later, in Oracle’s September earnings call, Larry did an about-face and announced that Oracle would release its own in-memory database before SAP could get to market.  He didn’t win that race: SAP launched HANA on Dec 1, 2010. 

The announcement explained that “SAP HANA is an integrated database and calculation layer that allows the processing of massive quantities of real-time data in main memory to provide immediate results from analyses and transactions.”  This is business at the speed of thought – “transactions + analytics” and not analytics in isolation.  A couple of recent tweets from Dr. Vishal Sikka, member of the SAP Executive Board, capture our direction:

vsikka: #HANA’s endeavor: serve the new real-time. On-the-fly analytics, planning on real-time data. Cutting layers, bridging divides, open systems.

vsikka: New real-time = OLTP & OLAP. Unlimited scale & Open systems. Text & Structured data. Non-disruptive Evolution & Unprecedented applications.

Since that time, we’ve begun delivering amazing new applications on top of SAP HANA. At SAPPHIRE NOW in May 2011 and again at SAP TechEd last month, customers and partners provided testimonials to the value of SAP HANA.  Going forward, HANA will become the next-generation platform that combines in-memory and cloud seamlessly and a “mobile-first” approach to developing apps.  Industry and technical influencers, customers, and partners have validated our approach.

On the other hand, let’s take a look at what Oracle has announced – 18 months after SAP.  Personally, I’m mostly underwhelmed for both technical and customer reasons:

Technical: I had expected Vishal would provide detailed technical analysis of Exalytics in his blog but, after he took a good look at Oracle’s new product, he had a “there is no there, there” reaction in this impromptu video.  As he tweeted the night of the announcement:

vsikka: Exalytics: 15 yr-old timesten + 15 yr-old essbase + sun hw = same old aggregates, layers, complexity & closed. Totally misses new real-time. 

The announcement suggests that Exadata needs Exalytics for analytics acceleration, which means Oracle feels like it must accelerate the accelerator.  SAP HANA is exactly opposite of that kind of afterthought: the HANA architecture is built ground-up for the new reality of real real-time applications. 

And how does Oracle support the totality of Big Data? If we take a look at the V’s of Big Data—volume, velocity, and variety—it looks like Oracle expects you to buy an appliance for each one: – Volume (Exadata), Velocity (Exalytics), and Variety (Exa-ToBeAnnounced). It’s ExaSperating.

Customer-centric: In attempting a retro-IBM style move with their hardware infatuation, Oracle has forgotten the customer.  I watched hours of videos and read litanies of blogs but I can’t find answers to simple questions:

  • How do you know which Exa appliance to use and when? 
  • How do you cost-effectively manage the growing litter of appliances or engineered systems?
  • How do you extend an “engineered system” that is not open or provide choice?
  • Where are the applications that run on these Exa engineered systems?

     

I suspect the answers are: it doesn’t matter; you want an all-Oracle optimized hardware stack.

 One more point of contrast.  SAP is about to offer HANA as a database under SAP NetWeaver Business Warehouse.   With that, we provide our customers a choice of whether they continue with relational databases, or whether they want to enable a real-time data warehouse by adopting SAP HANA.   Will Oracle offer that type of choice? Only, it would seem, if it’s willing to risk undercutting its core database franchise. 

To report this post you need to login first.

4 Comments

You must be Logged on to comment or reply to a post.

  1. Bala Prabahar
    Hi Aiaz,

    Based on what I learned on Sunday on ExaLytics -(I watched that video several times), you nailed ExaLytics with this statement: Oracle feels like it must accelerate the accelerator.

    Mr Ellison said “You can either connect ExaLytics to Exadata or to the existing server (I assume he meant Oracle 11g)”. If this is the case, then I would like to know why ExaLytics is better than BWA.

    We don’t need  Essbase or Heuristic Adaptive In_Memory Caching in BWA. BWA seems to be simpler than ExaLytics.

    Please note my knowledge on ExaLytics is limited so am looking for additional details.

    In addition, if anyone knows website discussing the internals of ExaLytics (like HPI for SAP technologies), please let me know.

    IMHO – with a limited knowledge on ExaLytics – ExaLytics versus HANA is not apples to apples comparison. My thoughts on HANA is not based on marketing hype but based on the videos I watched in Hasso Plattner Institute and http://www.experiencesaphana.com. Those videos discuss the internals of HANA.

    Best regards,
    Bala

    (0) 
  2. Alvaro Tejada Galindo
    Aiaz:

    I found pretty funny those comments from Larry…he first criticize SAP efforts and then Oracle tries (and I said “tries”) to do exactly the same thing…
    For me…he was just scared of seeing SAP doing the impossible…which is something that SAP does real well -;)

    I’m pretty sure that Oracle will try to speed up to keep up to date, which will result on messy solutions that would harder to take off that to install them -:)

    Greetings,
    Blag.

    (0) 
  3. Martin Lauer
    hello Aiaz,

    I’ not sure if exalytics does target SAP in the first place. it coudl also be that oracle has to react to the new players like qlikview or tibco spitfire.

    honestly I thnk that SAP and Oracle just started at different positions and will at some point reach the same goal: Oracle started to accelerate its disk-based DB and now adss in memory anlytics. SAP started within memory and (my assumption) will have to add disks (or flash or tape) as well in order to persist data (archiving!).

    i think there at least one technical difference why oracle needs to accerallerate its exa DB machine: latency.

    Sizing: the sizing of exalytics is most interesting for me: 40 cores that scan 1 TB (compressed) of RAM in 5 seconds. certainly to big for small companies (that should prefer MS PowerPivot then) but also small in my opinion because I thought that HANA supports much mores cores (esp. for complex computations).

    BTW: i don’t think that the software provided with exalytics hasn’t changed over the last 15 years :-). …some optimizations were mentioned

    (0) 

Leave a Reply