Home The Company Publications Products Links Tips Jobs

Experiences With The Event Replicator For ADABAS

Replicate ADABAS on the Mainframe to ADABAS on Windows

By Dieter W. Storr

Last update: 17 July 2010

Companies as well as state and federal government agencies try to save money and migrate their core applications from the mainframe and ADABAS to other hardware platforms and database systems. But why lose all your expertise in Natural and ADABAS and re-train your staff members just to be more ‘sexy?’

Replication is the buzzword and, of course, using the Event Replicator for ADABAS from Software AG, leaves your core applications on the mainframe, replicates the entire database to ADABAS on a Windows server and runs your batch processes faster and less expensive on a PC compared to a mainframe.

This presentation will discuss the basics of the Event Replicator and show how to set-up files to be replicated and create the definitions for destination, subscription, GFB, initial state and the new ADADBS function. Not all utility functions can currently be automatically replicated on a non-mainframe platform. File changes, for example, need special handling on the subscription, the replicator engine and the target site. An example will point out the process step by step.

During tests with production-sized data, for example 160 million records, communication problems occurred, parameters haven’t been calculated correctly and RPL buffers and SLOG overflowed. The initial-state procedure also seemed to be too time-consuming and other procedures had to be found.

Despite users’ statements that the replication works correctly and comparisons between source and target are not necessary, self-written comparison programs sometimes show differences. Two methods will be discussed how to compare source and target files to be sure that both are identical.

Last but not least, this presentation will also show the very important recovery function. Normally, the Event Replicator Server restarts automatically after an abnormal end and is able to recover any lost replication. But in some cases, data get lost and the replication must be replayed. The new ADARPL utility and its REPLAY function is similar to the known regenerate function of ADARES. Both ADARPL and ADARES need the protection log as input.

Presentation from June 2010 in Prague, Czech Republic
Updated version from 2009