Saturday, February 13, 2016

4.3 Corporate Data Pool becomes reality: what does it mean?

You can easily guess the picture when we do the procedure described in chapter 4.2 multiple times: we will end up with a single HANA database for many applications. From now on I will use the following picture to illustrate how a Corporate Data Pool on HANA works and what it means for our IT landscape.

The box on the bottom of the picture represents our corporate HANA server which we will use as single database for numerous applications. Three applications are shown in the picture for reasons of simplification but it could and will be many more. These applications can be anything, even a mobile app, but to evaluate the benefits we are interested in heavy-weight mission-critical applications with significant business logic.

For now we will assume that there exists a second HANA box for all the SAP systems. Technically it is not necessary but SAP will not allow us to take other applications in their dedicated database. Let’s not worry about it for now as this is not really an issue. We handled 500 databases before, we can handle two!

Let’s have a look how the landscape has changed and how we benefit from an IT perspective. I will talk about the business benefits in chapter 5.

Where did we come from? Let’s take the two Java applications from chapter 4.1 and 4.2 and put them in the context of a real system landscape, let’s say within the Aftersales department of a large discrete manufacturer. Here are typical parameters that I will use to describe what exactly has changed:
  • 1 IBM mainframe which contains the whole parts’ master and various programs (Cobol)
  • 15 productive custom systems implemented in 3 different programming languages (Java, C++, VisualBasic)
  • 15 productive relational databases from 3 vendors (Oracle, Microsoft, OpenSource)
  • 50 interfaces/adapters between systems; an integration infrastructure is in place but it is not connected to all systems and it only works in asynchronous batch mode
  • 3-tier landscape (development, integration/test, production)
  • 5 out of 15 systems are dedicated reporting systems (duplicated data, read-only)
  • The overall data volume in these 16 systems is around 35 TB
You do not need much imagination that all of this is very complex, expensive and that the data quality in the systems is very low. As described in chapter 4.2 we have migrated step-by-step all the databases to HANA which was the easiest part. We got rid of the duplicated data and implemented a basic data model with HANA views on top of it. The client-server applications in Java, C++ and Visual Basic ran pretty much out-of-the-box. Unfortunately the customer has used some database vendor-proprietary technologies which caused some extra efforts but which we could adapt with the help of intelligent tools. That was the same with the mainframe which now runs on the Corporate Data Pool as well. The Cobol programs were translated into Java applications.

Here is the list of benefits that we realized by doing all this:
  1. Shutdown of mainframe. This dramatically lowered our cost of operations from a hardware and a knowledge perspective as there were not many Cobol developers left
  2. The business logic of our 15 application continues to work like before but now at high speed with accurate and consistent data
  3. We shut down 45 database servers and replaced them with 3 HANA servers
  4. We shut down all of our 50 interfaces and decommissioned the integration software
  5. We decommissioned all 5 of our dedicated reporting systems as we do not need them anymore. We are able to do reporting on transactional data at any time in any granularity.
  6. The overall data volume decreased to slightly below 4 TB. We run the whole data pool on a single scale-up machine. For further data growth we will implement the “dynamic data tiering concept” of SAP HANA as we do not need all of our historical data in main memory. This will again decrease our memory consumption by another 50-70%.
  7. We decommissioned 80% of our backup capacity. More than that we are now able to make a consistent backup of all our business data and restore it to any point in time if we needed to.
The benefits for IT are huge. And all that we did was basically exchanging the database.

No comments:

Post a Comment