Database-free SAP?

By Sam Sliman
President, Optimal Solutions Integration, Inc.

Can you run SAP without a database? The question is not as far-fetched as one might suppose, particularly given recent advances in in-memory technology, the precipitous drop in RAM cost, and the fact that SAP would relish the opportunity to pinch off the revenue many (perhaps most) of its customers generate for Oracle by running SAP on an Oracle database.

OK, so maybe SAP will never be completely database free--certain functions and business continuity considerations might always require some data to be stored in a database--but relational database management systems (RDBMS) as we know them may be headed for extinction.

The reasons for the demise of the RDBMS are numerous and varied, but relative cost and an inability to perform fast analysis on massive amounts of data top the list. Add to this mounting market forces that favor in-memory technology, including SAP’s vested interest in seeing this technology take off, and it’s not so hard to envision a world where SAP runs “database-free.”

Database cost, complexity and limited capability

Databases and their massive disk farms can be costly to maintain. Moreover, they are CPU hogs that pose daunting power and cooling challenges. According to Gartner, by 2008, nearly 50% of data centers worldwide will lack the necessary power and cooling capacity to support high-density equipment. This problem will most likely worsen. According to IDC, 30% of companies with $500 million or more in revenue expect their data warehouses to grow at least 100% over the next three years.

At best, the RDBMS is a necessary evil. While good at storing persistent, structured data, RDBMSs are challenged when used for business intelligence purposes; that is, for when massive quantities of unstructured data must be quickly analyzed in order to inform critical, timely business decisions. When large amounts of data need to be analyzed efficiently and quickly, querying an operational database takes too long and consumes too many computing resources.

Advances in in-memory technology

With all the attention recently lavished on the wonders of Web 2.0, not much mindshare has been afforded lately to the inexorable march of Moore’s law, particularly as it pertains to the ever-decreasing per-unit memory costs. The average price for 1 gigabyte of random-access memory (RAM) has fallen from more than $1,000 in 2000 to slightly more than $100 in 2005. Gartner estimates that 1 gigabyte of RAM will drop to less than $10 by 2012.

The advent of 64-bit chips is also spurring adoption of in-memory technologies. Servers with multiple 64-bit Intel CPUs routinely handle 100 gigabytes of RAM and more.

In-memory excels when there is a need for speed. A random byte of data can be found up to 1 million times faster in RAM than in a traditional disk-based data warehouse. What’s more, in-memory technology not only retrieves the data faster than traditional aggregate-based architectures, but it also performs calculations on the query results much faster than disk-based approaches.

It seems inevitable that with time and by virtue of ever-increasing processing power and ever-decreasing RAM cost, in-memory technology at the application level will leave the slow-spinning disks of the traditional RDBMS in the dust.

SAP’s in-memory foray

SAP’s ‘hands-on’ history with in-memory technology, beginning with SAP liveCache technology, pre-dates that of competitors such as Oracle, who bought its way into the in-memory market by acquiring TimesTen in 2005.

SAP liveCache combines the benefits of relational and object-oriented data modeling in an in-memory database system, which presently powers SAP Advanced Planning and Optimization (SAP APO), the planning solution within mySAP SCM.

In 2006, SAP introduced its Business Intelligence Accelerator (SAP BIA), an analytic engine for NetWeaver that utilizes in-memory technology to enable very fast query processing—and has no requirements for an underlying database. Essentially an appliance, the BI Accelerator bundles hardware and software in a single package. To create the BI accelerator appliance, SAP has partnered with Intel, which provides the processors, and HP and IBM, which provide their respective server and storage technologies.

According to IDC, the BI Accelerator, in essence, is a highly scalable analytic server that processes queries initiated by users of SAP NetWeaver BI. SAP's TREX search technology in conjunction with blade server architecture provided by its hardware partners make SAP BIA unique and highly scalable.

According to Gartner, by 2012 70% of Global 1000 organizations will load detailed data into memory as the primary method to optimize BI application performance. Whether a similar approach can address the needs of operational transactions is up for debate but, in theory, nothing insurmountable stands in the way. Soon, running a wide range of mission-critical applications entirely within memory – and without an external database –will not only be a viable option, but quite possibly the preferable one as well.