Hi
I searched in google to find usage of virtual memory and Swap in InfoSphere but couldn't find much information.
and also I want to know what is the rational behind allocation for these two(How much need to be allocated and on what basis)
Can someone give some details on it.
Thanks
Naga
Virtual Memory v Swap
Moderators: chulett, rschirm, roy
That's because both virtual memory and swap are controlled by the operating system. If you are looking for guidelines for usage, there are none. That's because DataStage jobs vary so widely in terms of design and memory usage.
I have one client that has thousands of tiny jobs that move very small amounts of data very quickly. They can have dozens, if not hundreds of jobs active at any given time, and work quite effectively with an average system (in terms of memory).
I have another client with jobs that load large lookup tables into memory and process millions of rows. That client needed all the memory they could stuff into the machine to support their particular architecture and improve performance.
The only way to make any reasonable estimates is to look at an existing system and see how memory and swap are currently being utilized by the current workload. If you monitor it over time, you'll see peak utilization and can typically identify specific job designs that are driving high memory utilization. If memory usage is too high, then those jobs must be optimized or re-designed. That or add more memory to reduce swap.
I have one client that has thousands of tiny jobs that move very small amounts of data very quickly. They can have dozens, if not hundreds of jobs active at any given time, and work quite effectively with an average system (in terms of memory).
I have another client with jobs that load large lookup tables into memory and process millions of rows. That client needed all the memory they could stuff into the machine to support their particular architecture and improve performance.
The only way to make any reasonable estimates is to look at an existing system and see how memory and swap are currently being utilized by the current workload. If you monitor it over time, you'll see peak utilization and can typically identify specific job designs that are driving high memory utilization. If memory usage is too high, then those jobs must be optimized or re-designed. That or add more memory to reduce swap.
