jump to navigation

SAP ERP and SAP HANA April 19, 2012

Posted by Mich Talebzadeh in SAP.

SAP ERP – SAP Enterprise Resource Planning  (started as an invoicing/ledger system) comes in different flavours including one for Small to Medium Enterprises (SME). It has been in operation for many years. For backend it can use its own propriety database called  MaxDB, that SAP ships. However, customer may prefer to use another database such as Oracle or Sybase ASE. SAP will be phasing out MaxDB and replacing that with Sybase ASE.

As I understand, as of now the way it achieves this so called data agnosticism is by sticking to portability aspects of SQL. SAP ERP uses a high level programming language called ABAP http://en.wikipedia.org/wiki/ABAP . There is a layer in Netweaver (the engine that executes ABAP) which translates the data access portion of ABAP into pure SQL and executes this against whatever database is serving up as data store. So that translation could be for transact SQL or Oracle SQL or PL/SQL. To achieve this coupling, the database itself does not use any stored procedure or anything proprietary,  so it makes it easier for SAP to install the server/database itself as customer wants (Oracle/ASE etc) as part of SAP ERP package. The traditional design is based on 2-tier architecture but this could be 3-tier of course (the database running on another host).

SAP – HANA – SAP (High Performance Analytic Appliance) is a combined hardware and software kit. SAP HANA enables immediate access, model and analyze all of the transactional and analytical data *in memory* from multiple sources on this appliance/kit.

To achieve this SAP uses hardware from HP and probably others and uses the large amount of memory available  to create an in-memory analytics data store. The technology used deploys columnar database (similar to Sybase IQ), compression and the usual parallel processing. This in effect allows SAP to provide up-to-date report in near real-time mode to the users by eliminating disk usage. For hardware it uses HP ProLiant DL580 G7 and DL980 G7 servers that provides up to 2TB of memory in a single server. To achieve the work load HANA deploys an in-memory computing engine and a data calculation engine.

As I mentionedabove, HANA will act as the in-memory repository for many SAP products including feeds from SAP ERP, SAP Business Information Warehouse (BW) and potentially from any other source. If you look at HANA technical white papers from https://www.experiencesaphana.com/community/blogs/blog/2012/04/12/sap-hana-technical-overview-an-entry-point-to-our-revolutionary-chapter ,you will see that for real time replication service say from SAP ERP databases (log based ones) it will use Sybase replication server. However, IMO Sybase replication server needs to go faster in order to deliver data to HANA IMDB. For example it needs to eliminate or replace all the disk based stable devices. HANA IMDB is also tuned for SAP Business Objects BI solutions. Now where does Sybase IQ fit here? Well if needed data can be flushed out from IMDB to Sybase IQ for persistent storage as an option. That could be on the roadmap.

HANA can handle both row and columnar data and it is indeed very interesting to have an optimizer that can handle row based implementation of relation model (row store) plus column based implementation of relation model (column store).

At this time I don’t have much of a view on HANA as I am still very much in learning mode regarding what it really is and what it really can do. In many ways it seems similar to Oracle Exalytics, though there are also significant differences. For example, HANA is currently only offered as an appliance, while Oracle TimesTen IMDB (and EssBase) can be purchased and used on regular hardware (though the TimesTen Exalytics specific features can only be used on Oracle Exalytics). Also, at least at the moment, it seems that SAP are mostly positioning HANA to work with other SAP products not as a more general purpose thing. Oracle Exalytics on the other hand can inter-operate with many different data sources from different vendors and Oracle TimesTen is also pretty open in that regard.


No comments yet — be the first.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: