Hi,
We follow a standard 3 Tier QVD Layered Architecture.The current combined size of all QVDs is around 12GB.
Towards the end we have a step whereby we concatenate all the QVDs to create a STAR like data model.
We are observing that at the step where we concatenate all the QVDs, the memory usage is as high as 60GB for a period of 2 hours.
Admins have taken the drastic step of killing that job because of so high memory usage.
The Prod server is a clustered 2 machine setup with 64GB RAM each.
Please let me know from your project experiences:-
1) Is it normal for data contained in 12 GB QVD to use upto 60GB RAM while merging the QVDs.
2) What ways we can explore to bring down the memroy usage.
3) Or should we increase the RAM capacity in the Servers.
Cheers,
Anindya
