SAP Betting ‘Main Street Businesses’ Won’t Embrace Hadoop for Big Data

Loraine Lawson

If disk is the problem, get disk out of the equation. Right? Don’t try and optimize for the disk, just remove it completely. Obviously SAP HANA has disk capabilities. We store our log on the disk. We back up to a disk, but fundamentally all the operations happen in-memory. And so we get around that disk bottleneck.

So that’s why we believe in-memory is at the core of a company’s Big Data strategy, especially as you look to build that into business processes.

When you start talking about, “I want to do real-time analysis inside my business process, within my ERP system,” what you're really talking about is massive simplification of an IT architecture so you can analyze and transact in real time at the same time. The only way to do that is with an in-memory database, so that is also fundamental to our strategy.

We’re setting the stage to be able to do real-time Big Data analytics inside your enterprise applications. HANA is the centerpiece and we have lots of pieces around it.

Then our strategy is to build out applications, so we have ballpark about 25 to 30 applications today that are HANA-enabled. Many of them have Big Data capabilities built right into that. You can store large amounts of data and analyze that within the application.

We’re also enabling our network of partners with their own applications to run on the HANA platform and to leverage whatever data they’ve got, whether it’s data stored in an SAP application or whether it’s stored in other systems to be able to leverage that in different ways.

Lawson: So for existing SAP, like if you run an SAP ERP system, to take advantage of that, that’s going to need more than a software upgrade? That’s going to be investing in HANA? Does that change the pricing for your ERP systems significantly to do that?

Jonker: So I will defer pricing questions to someone else, but generally the answer is the way our customers buy our ERP systems is that there is no dramatic cost to run ERP on HANA.*

It’s not some sort of dramatic cost to that. Now, there may be cost for other applications that leverage some of those insights, but overall, there’s going to be a reduction in cost.

The reason is that a lot of the costs enterprises face today are with the complexity of their IT environment. Enterprises have hobbled together these very complex environments, because they're running on databases that were designed for the 20th century.

Let’s take as an example: You have your ERP application. You run your ERP application on a traditional relational database, say it’s Oracle, maybe it’s IBM DB2, whatever it is.  Now, you want to analyze that data as part of your business process, right? So what companies have done is they’ve created a separate reporting database so that they can just do basic reporting on that application data. And then they actually need to do some reporting from data and other systems, so they create a data warehouse.

Now that data warehouse is built on a completely different database again, because the traditional relational database is good at transactions, but the data warehouse needs to be good at analysis. So a lot of them pick other databases, like, say, Teradata. But then they need to give too many people access to the Teradata system and Teradata can’t handle all these users hitting it concurrently, so they create all these extra databases with data marts, and then for some reason the data marts aren’t performing well enough, so they're creating aggregates of the data. They're simplifying the data. They're creating cubes. We do OLAP or aggregates of data because if we run against all the detailed information, things take too long, i.e., the database isn’t fast enough.

The real cost here is all that complexity. Think of all the systems we’ve just introduced in order to go from transacting to analyzing. When you run in memory, when you get the disk out of the equation, you can collapse all that into one platform. You can run your transactions and your analysis on one system against all the detailed information.

There’s no need to create all these aggregates or these cubes or all the summary of data.  There’s no need to throw data away. You can keep it all there in one system and so the cost of a solution like this goes down, in that the overall complexity is greatly simplified with a solution like this.

*SAP ERP built on HANA is a priced as a percentage of the SAP ERP license cost, with buyers able to choose from seven pre-approved hardware vendors, who offer different hardware configurations optimized for HANA. The box is shipped with HANA installed.

Add Comment      Leave a comment on this blog post
May 9, 2013 9:32 AM David Jonker David Jonker  says:
SAP fully expects Mainstreet will adopt Hadoop solutions and have a strategy to ensure it tightly integrates into a complete Big Data solution. We believe SAP HANA and Hadoop have different, complementary roles to play. Please check out this blog post that further articulates SAP's Hadoop strategy: Reply
May 9, 2013 5:43 PM Charles Chen Charles Chen  says:
Maybe, managers can proceed free conversation with SAP on business strategical topic. SAP may become strategic robot or manager's aide. Reply

Post a comment





(Maximum characters: 1200). You have 1200 characters left.



Subscribe to our Newsletters

Sign up now and get the best business technology insights direct to your inbox.