SAP Sapphire: Hasso Plattner talks in-memory computing, slams Oracle

SAP Sapphire: Hasso Plattner talks in-memory computing, slams Oracle

Summary: Legendary SAP co-founder Hasso Plattner discusses the limits of today's technology and why in-memory computing will change everything -- nevermind what Larry Ellison says about it.


ORLANDO, Fla. -- Hasso Plattner, one of the five co-founders of SAP and chairman of its supervisory board, took to the stage here at the company's Sapphire Now conference to clarify the potential that HANA, the company's in-memory database technology, will have on the company's future -- from both technical and strategic standpoints.

"HANA is not an experimental thing," he said. "We got lucky and we found something and it runs things 10, 100, 1,000 times faster. HANA is a platform and is on the way to become the foundation for most of our applications, whether they're on-premise or in the cloud."

See also: SAPPHIRE Now day 2 wrap - a cloudy and confusing daySuccessFactors' Lars Dalgaard swings for the cloud fencesCEO outlines vision for 'a better future'SAP HANA teams with Opera Solutions for real-time packaged analyticsSustainability is dead. Long live sustainability

Plattner wasted no time diving deep into the technical details of the technology, a somewhat odd choice given the thousands of businesspeople in the audience. Nonetheless, he was frank about the limitations of today's technology and why he thinks HANA will help free companies from them.

His talk was easily more than an hour long, and largely involved explaining the benefits of HANA by walking through it step-by-step. Still, he offered several observations about the technology and its relevance along the way.

A collection of highlights:

  • "I'm not saying aggregates are forbidden, but aggregates are not the organization of choice."
  • Don't expect a single-select database to exceed the physical capabilities of a computer.
  • "We have to make changes to the code. And we cannot abandon 99 percent of our customers" who are sitting on [IBM] DB2, Oracle, NextDB, Microsoft, etc. "We will not have a version, from a functionality point of view, completely different" between HANA and the existing platforms customers use and from which SAP makes money.
  • Customers will receive benefits of the suite, regardless of the setup. There will not be HANAs and HANA-nots.
  • "I feel very good about this, that SAP is not continuing to grind and hone what we did 15 years ago."
  • And with the rise of mobile, in-memory matters even more. You're not going to wait more than three seconds for something. "This is where the world is going: direct access from anywhere to any type of data, whether it's a small piece of data or a small answer but a long algorithm to create that answer. The user doesn't care about this."
  • "New applications will have to deal with big data. We have to analyze it on the fly, so we have to have a system that is transactional and analytical at the same time. We cannot have a multi-stage system. This is too slow for modern applications."
  • "SQL is not dead. The world is living on SQL. If you pull SQL out of the enterprise world, the world will break down."
  • "We run 18 times faster today in the data transfer. Why don't we run 100 times faster? We still have to write to disk. This is the bottleneck. We cannot be 500 times faster in data input."
  • "Business reporting is not dealing with objects, it is dealing with relationships between objects."
  • "There are many uncharted areas where we can go now and at least provide technology, probably help to build applications or let others build applications. HANA is totally open to third parties. This will be the difference. We tried this 20 years ago [but the platform was dropped in the face of another success]…this time we will not make this strategic decision again to go for SAP applications only. We go for non-SAP applications as much as we go for our own."
  • "We find more and more companies where the application of speed opens up the door for completely new applications."
  • "We have to deal with petabytes in the future. We cannot deal with this reorganization we've done in the last 20 years."

Plattner didn't resist in taking a few swipes at rival Larry Ellison of Oracle out for spreading fear, uncertainty and doubt about a technology with implications not yet fully understood.

"Please, please, please, my dear competition," Plattner said, pleading, "We can beat each other and fight each other as much as we want and argue, but do not predict how a system really works when you really don't know and don't want to know. Either be better informed, or don't do it."

At one point, Plattner pointed to the massive display behind him, projecting a diagram of the HANA system, to ensure that it was visible via streaming video "for our friends in Redwood Shores."

He added, later: "We should not let FUD creep into our core operation. The other ones don't know enough."

But back to the topic at hand. Plattner employed several technology demonstrations during his talk to illustrate his claims. Among them: SAP's Business One ERP system; the use of real-time promotions in a Bigpoint video game; an overview of luxury outfitter Burberry's clientelling app for sales associates by CTO John L. Douglas; and a demonstration of HANA's high-availability capabilities by systematically turning off servers at its datacenter in Santa Clara, Calif. and watching the system respond in real time.

At the core of all of it? In-memory computing.

"It is fantastic for me to see after 44 years in the industry, that we have computer resources of that magnitude for relatively little money."

But he couldn't resist one more dig.

"Oracle BI runs nicely on HANA," Plattner said, a twinkle in his eye. "[But] it's not the preferred version from an Oracle perspective."

Photo: Tom Raftery

Topics: Oracle, SAP

Andrew Nusca

About Andrew Nusca

Andrew Nusca is a former writer-editor for ZDNet and contributor to CNET. During his tenure, he was the editor of SmartPlanet, ZDNet's sister site about innovation.

Kick off your day with ZDNet's daily email newsletter. It's the freshest tech news and opinion, served hot. Get it.


Log in or register to join the discussion
  • But surely for mobile

    the biggest bottleneck is the network, not the disk on the server?

    I definitely agree about not splitting operational and analytical systems and that relational is not going to go away (SQL maybe, but relational definitely not).
  • Another fail for Microsoft.

    MS and their ERP warez are dead in the water. No stimulation in MS ERP tech at all. Just old crap refreshed into MS warez.

    Good bye Dynamics division! Another market you failed to conquer despite your money infusion and limited mind-set.
    The Danger is Microsoft
    • I get it. Downvote because you want to support Microsoft

      but don't reply because you would look like a donkey (err...or the other word for that animal).
      The Danger is Microsoft
      • It's mainly because most people can't be bothered with you

        I'm waiting to run some code, so I thought I'd fill you in.
    • Warez?

      Isn't that the other side of the Rio Grande from El Paso?
  • HANA is very good

    SAP is doing very good work with HANA. I have tried the ICE which is the beta engine and it is fast for analytics on business one. I thought MS had a better product for SMEs as in the past few years the bugs in business one was just anoying and it was missing alot of functionality. Hopefully when HANA is being sold with B1 licenses that the new interactive analysis will compete against NAVs good reporting tools.
    It would not surprise me if many of Oracles ERP customers adopt HANA and something like SAP BW as there are alot of companies with petrabytes of data which is being archived and not used for analytics.
  • I suspect the in-memory thing is a little overplayed

    All the major SQL-DBMS vendors have steadily changed their products to do more in memory. Optimizers have been modified to take advantage of larger memory capacity, caching algorithms have been improved, and so on.

    The direction is clearly towards more being done in memory, but I think it is an evolution rather than a revolution and you don't necessarily need a whole new SQL-DBMS to do it.

    On the other hand SAP are absolutely right in going all out for relational and ignoring the other options.
  • will companies beable to keep data safe with in memory techniques?

    Malware can be designed to attack memory, application buffers, caches and such. Will data running in memory be encrypted and wouln't this create a processing over head for the database funtions it's self.
    • Huh? Apps run in memory, and without memory apps will not run...

      What are you talking about? Malware is malware, protect yourself.
  • In-memory - and memory choices

    To deal with the volatile nature of memory - just use the upcoming DDR3 NVvault type of products from Netlist. Back up memory to flash built into the memory module on power failure.

    And you will need a "load-reduction" solution to support memory at 3 DPC (and at 2 DPC and 3 DPC when 32GB RDIMM 4-rank roll around) - like Netlist's HyperCloud (available at IBM as HCDIMM and at HP as HDIMM).

    Then you can have 768GB (populating all 24 DIMM slots on a 2-socket server) - and put all your databases in memory.