job recovery using control tables in Dataservices

[former_member646254]

Hi, can any one help to know the step by step process to recover job using control tables only in Dataservices. If my job failed in middle when loading huge volume of data...??

2.what is the difference between recover from last execution option and control tables when job is failed in middle of execution due to any reason..??

3.how to create control tables..?? To track job status, time stamp, job run id, volume of records it's processed.(count of records).

4.how do we come to know exactly how many records processed and at what point of record job was failed and how to recovery and all..??

[werner_daehn]

I wrote an article about that a while back

https://wiki.scn.sap.com/wiki/display/EIM/Recovery

To cut a long story short, recovery within a dataflow cannot be done. At restart you would need to recreate the exact same condition, guarantee that the same data is read in the same order, all buffers and internal caches etc. If the dataflow would write that also, then the loading would slow down by factors. So instead of a non-recoverable load finished in 30 minutes you would be able to recover it but it runs for one hour longer. And even then you don't know if the source provides the same data in the same order.

For such cases it is better to split the load into segments and call the dataflow multiple times. E.g. load one financial year after the other. Then the overhead is acceptable, the restart can skip the years already loaded and has a clear starting point.

The other questions should be answered with above link. As a general rule, while the error handling and monitoring is not perfect, recreating a similar logic manually is something I would not advise. There is so much information in the DS repository etc. So make sure you know that area before you really decide to build something yourself.