Skip to Content

How to reasonably split an at-once datasource replication process made for a whole source system

A resource problem may occur, when all datasources of a single source system must be replicated in one step if there is a high number of datasources (e.g. several thousand) together with a high number of languages (e.g. more than ten) that are requested in the replication. In case it is necessary, the replication for single datasources can be done any time to obtain the complete list of languages by replicating them individually from the context menu of the datasource tree. However, I would like to handle the situation when all datasources need to be replicated at once for a specific source system.  Background of the problem: currently there is no option available which performs a “packaged” replication instead of the “at-once” replication, as this depends from the source system’s PI_BASIS as well. We hope to have this feature in coming releases, until that time we may consider the following options.   1) Add some memory The fastest way to solve the problem would be to enlarge the memory size which could be allocated by one process. This could be set independently from the available physical memory. However, it is not a nice solution so lets just skip it.   2) Splitting replication by node hierarchy Replicate the hierarchy nodes with the button “Replicate Tree Metadata” in the source system sight. image  Then replicate the datasources contained with right-click on the first level nodes choosing “replicate metadata”. image  This method doesnt require any technical knowledge, however, it may cost quite some time. But if we know the processes behind the replication, we can make it much faster. So why not debugging a little bit?   3) Splitting replication by languages With a little debugging, the high amount of datasources for a source system can be replicated in two or three batches by dividing the required list of languages.  Here is the how-to:  Call transaction SE80 and set a breakpoint in the Include LRSAOSF04 and Form METADATA_20_UPLOAD at the following statement: PERFORM languages_get USING i_t_langu CHANGING l_t_langu.image  Then start the replication for the whole source system in the Data Warehousing Workbench. When reaching the breakpoint, press F6 once, so that the internal table L_T_LANGU gets filled. image  Now reduce the content of the internal table to the required parts.  image  Finally you can exit the debugger and switch back to replication. image  After successful replication of the first batch of languages you can repeat the above steps for the remaining languages.
Be the first to leave a comment
You must be Logged on to comment or reply to a post.