SAP HANA Platform

SAP's long-standing strategy was to provide an application layer independent of the database. Accordingly, SAP's focus in the past was not on developing its own databases, and with SAP MaxDB, the SAP product portfolio contained only one database that could hardly compete with the products of specialized database manufacturers. This situation has changed dramatically with the SAP HANA platform introduced by SAP in 2010. With SAP HANA, more and more functionalities that were previously located in the application layer have been and are being moved to the database. On the one hand, of course, the independence of the application layer from the database layer is lost, but the advantages listed below more than make up for this disadvantage.

SAP HANA meets the Atomicity, Consistency, Isolation, and Durability requirements (ACID standards) and the SQL 92 standard. It is an in-memory column-oriented database. This optimizes the advantages of multicore processing and single-instruction (SIMD). SAP HANA is column-oriented by default but also enables the so-called row store, which decomposes and stores database tables in a sequence of rows. This makes the table easier to restore and writing new rows is a relatively simple operation in this structure.

Database tables in SAP HANA are stored in a sequence of columns by default, which is advantageous for in-memory multi-core parallel processing. The data here is prepartitioned in independent data blocks that can be processed directly in multiple cores of the processor simultaneously. Another advantage of column orientation is that it does not require synthetic analytic indexes, since the sequence order is already a natural index. It enables efficient compression, e.g. by Dictionary Compression. With in-memory column store, you can perform on-the-fly aggregations, eliminating pre-aggregates and materialized views.

In principle, the amount of data in SAP HANA is not limited by main memory space, because the architecture allows data to be stored not only as in-memory tables, but also as multistore and disk tables. Management, encryption, backup and recovery functions are equally guaranteed for all tables. Data can be merged when retrieving from the in-memory, multistore and disk tables. The memory location of a table can be changed at any time.

Data from other data sources such as Apache Hadoop and Apache Spark are also accessible via queries without having to transport them at great expense. Features such as SAP SQL Anywhere Suite embedded in remote devices enables remote data synchronization.

Since 2014, SAP HANA has been offered in a cloud installation as well as in a hybrid environment in addition to the classic on-premise installation.

In addition to ABAP, developers can also program in many other languages, including Java, JavaScript, Python, C++, Go, Node.JS, git, github, HTML5. Furthermore, UI libraries and semi-structured data formats such as JSON and XML are supported. The development environment in SAP HANA Studio is based on Eclipse. SAP HANA also integrates TensorFlow and R Server for data stream-oriented programming, statistical pattern recognition and business process automation.

Kreps, J. (July 31, 2014). O'REILLY. Retrieved from https://www.oreilly.com/ideas/why-local-state-is-a-fundamental-primitive-in-stream-processing

SAP SE. (20. 09 2018). SAP HANA In-Memory Data Platform. Retrieved from https://www.sap.com/products/hana.html