SAP HANA is the synthesis of three separate products – TREX, P*Time and MaxDB.
- TREX (Text Retrieval and Extraction) is a search engine. It began in 1996 as a student project at SAP in collaboration with DFKI. TREX became a standard component in SAP NetWeaver in 2000. In-memory attributes were added in 2002 and columnar data store was added in 2003, both as ways to enhance performance.
- In 2005 SAP acquired Menlo Park based Transact in Memory, Inc. With the acquisition came P*Time, an in-memory light-weight online transaction processing (OLTP) RDBMStechnology with a row-based data store.
- MaxDB (formerly SAP DB), a relational database coming from Nixdorf via Software AG (Adabas D) to SAP, was added to TREX and P*Time to provide persistence and more traditional database features like backup.
SAP HANA is not SAP’s first in-memory product. Business Warehouse Accelerator (BWA, formerly termed BIA) was designed to accelerate queries by storing BW infocubes in memory. This was followed in 2009 by Explorer Accelerated where SAP combined the Explorer BI tool with BWA as a tool for performing ad-hoc analyses. Other SAP products using in-memory technology wereCRM Segmentation, By Design (for analytics) and Enterprise Search (for role based search on structured and unstructured data). All of these were based on the TREX engine.
Taking a different approach Advanced Planning and Optimization (APO) used LiveCache for its analytics.
Big data refers to datasets that exceed the abilities of commonly used tools. While no formal definition based on size exists, these datasets typically reach terabytes (TB), petabytes (PB), or even exabytes in size. SAP has positioned HANA as its solution to big data challenges at the low end of this scale.At launch HANA started with 1TB of RAM supporting up to 5TB of uncompressed data. In late 2011 hardware with 8TB of RAM became available which supported up to 40TB of uncompressed data. SAP owned Sybase IQ with its more mature MapReduce-like functionality has been cited as a potentially better fit for larger datasets. By May 2012, HANA was able to run on servers with 100TB main memory powered by IBM. Hasso Plattner claimed that the system was big enough to run 8 largest SAP customers.