SAP Puts In-Memory Database At The Heart Of Its BusinessSuite ERP

January 14, 2013 Alex Woodie

SAP last week made the major leap of officially supporting the use of its HANA in-memory database to power SAP BusinessSuite, its flagship enterprise software. Putting HANA in charge of managing analytical and transactional applications and their data simultaneously marks a huge architectural change for the German ERP giant, and it shows that the company is not afraid to take chances. Breaking the barrier between transactional and analytical data will pay off, SAP says, because it will allow businesses to “analyze and predict instantly.”

Ever since IBM invented it and Oracle popularized it 30-plus years ago, the relational database management system (RDBMS) has been the go-to technology to power ERP and other enterprise applications that are the systems of record for businesses. The RDBMS has by no means been perfect, but it is excellent at processing online transactional processing (OLTP) workloads.

The main drawback with the RDBMS is that it is not real good at serving online analytic processing (OLAP) workloads. Sure, it can pull up reports and make sense of the data you already have. But if you want to get serious and mix tables and go into more than two dimensions–that is, to create new information and glean new insights from the data you already have through OLAP–then an RDBMS will not be your first choice of machine. You will want one of the more exotic multi-dimensional databases available on the market.

However, there are consequences to maintaining separate systems for production and analytic workloads. First, it is really expensive to build and maintain separate systems. Second, keeping data in synch between the production database and the data warehouse is a giant pain in the tootsie. Finally, the data in the warehouse is never as fresh as the data in the ERP system. For some companies in some industries, that delay is quite costly.

In 2008, SAP began working on a new database architecture called HANA that could be used for both transaction and analytical workloads. HANA (which doesn’t stand for anything anymore) is a combination of three separate products, including the TREX (Text Retrieval and Extraction) search engine, an in-memory RDBMS database for OLTP workloads called P*Time, and MaxDB, an established RDBMS that SAP bought from Software AG.

When HANA became generally available in 2012, it was only supported for analytic workloads, where it would run as a side-mart, bolt-on to a production system. SAP promised to one day run its core ERP applications on the HANA database. That day was last Thursday, when SAP announced that three out of four component of its BusinessSuite, including SAP ERP (enterprise resource planning), SAP CRM (customer relationship management), and SAP SCM (supply chain management) are now supported running on HANA. The fourth BusinessSuite component, ERP SRM (supplier relationship management) will be supported on HANA soon, SAP says, as will SAP Business All-In-One, the pre-configured version of BusinessSuite designed for smaller businesses.

There are no code differences between standard BusinessSuite configuration and the one powered by HANA. HANA is a fully certified database, joining other certified databases like Oracle, SQL Server, DB2, and Sybase (which is also now owned by SAP). SAP has put together a series of “rapid deployment” solutions to help speed customer migrations to BusinessSuite powered by HANA. The solutions include software and services, and can typically be completed in a manner of months, says Ken Tsai, vice president of SAP HANA product marketing.

SAP is trying to keep the pain of moving to BusinessSuite powered by HANA minimal, while at the same time instituting major architectural changes. In effect, SAP is moving from a three-tier system (after all, R/3, which is BusinessSuite’s previous name, basically stood for three tiers) to a much flatter two-tier system–or more correctly, to a two-and-a-half tier system, where the application server has been gutted and basically now just exists to enforce security. It is almost back all the way to the mainframe-based R2 suite.

“If you look underneath at the optimization techniques that BusinessSuite has gone through, we’ve pushed down data-intensive processes. Instead of writing them in ABAP, we write them into code directly into HANA,” Tsai says. “We looked at the application code, and [decided] that it makes sense not to put it in the ABAP tier any more, but to put it into the database tier, and then deliver it as stored procedures or SQL script codes, or deliver it as pre-delivered business function libraries in the SAP HANA platform. There are multiple optimization techniques that our developers have gone through.”

Enterprise architects can argue about what makes the best enterprise architecture until the cows come home. That’s why they get the big bucks. But what really matters to customers is results. Behind all the techno blather, business owners need simple and good reasons for ripping out their tried and trued RDBMS and putting all of their data eggs into the same HANA basket. The Germans understand this need, and that is why they invited John Deere, an early HANA adopter, to participate on opening day.

According to Derek Dyer, Deere’s director of global SAP services, putting HANA in charge of OLTP means that it can complete its materials requirement planning in real time. The company is also looking at implementing predictive maintenance for its equipment in the field. “This move will be a game-changer for John Deere,” Dyer says.

Other enterprise workloads that SAP sees benefiting from HANA’s in-memory capabilities include: including marketing analysis, financial close, receivables management, and consumer and social sentiment analysis. There’s a lot of data being generated here, and businesses are very eager to squeeze these data sets for nuggets of actionable data.

Predictive analytics is one of the areas of biggest potential gain for customers for agreeing to turn HANA loose on their OLTP data. SAP put a lot of new predictive capabilities into a new release of its CRM app last year, and it’s safe to say that SAP will be looking to add more predictive capabilities to its enterprise applications in the future.

All this probably won’t help sales of SAP’s BusinessObjects software. If this experiment succeeds, and SAP customers move en masse to putting HANA in charge of OLTP and analytic data and putting analytics directly into the enterprise products, then sales of SAP’s data warehouse software will also begin to dry up.

Shifting revenue from one product bucket to another is no big deal for SAP, as long as it stays in the family. But for Oracle, the change is potentially hurtful. SAP in the last few days has been fond of touting its status as the biggest reseller of Oracle’s database. It may not be any more.