Research

SAP HANA is a modern real-time database that combines advantage of traditional SQL databases with some innovations in the NoSQL area, in particular data compression, as with Google’s “Bigtable”. The compression allows to evaluate large amounts of data “in-memory”, i.e. is much faster without disk access.

But it does not work with any amount of data. We tested it together with SAP in a joint project on HANA extensively: The detailed XETRA overall market data required zip-compressed on average 250 MB per trading day and are stored in our own Hadoop infrastructure. A HANA with 70 GB of memory, was capable to store about 120 days of XETRA data or 30 GB of compressed data to evaluate it in memory. A HANA with 130 GB of storage made up to twice as much availalbe, i.e. one year of XETRA history (trades, no quotes) or compressed 60 GB of data.

We had heard Project Vora by the end of 2014, and are grateful, that the integration of Hadoop and HANA is available as an SAP-supported architecture and starts to find friends.

The benefits of HANA infrastructure are enormous, there are many old and new tools available, last but not least BI applications and ways of migration from the old SAP R/3 infrastructure to the new “S/4” (HANA). We were able to integrate our risk management libraries in large parts into the HANA kernel. Vora now proves that SAP eliminates the last weak points successively. We will try it.