What is the performance impact when I compile data collection?

[feedthemachine]

Unfortunately, I have not found this information available. Thus the question is: is there any impact during data collection? Let's say if I collect data using System Reporter or send constant queries to SSMC or run some CLI "show*" and "stat*" commands simultaneously, can I expect my performance to be affected somehow?

If so, what measures can I take to have this impact minimized (e.g. run collection off-peak hours, make queries more "oriented" in terms of what I want to retrieve, etc)?

Thank you!

[Dennis Handly]

It all depends.  How many nodes, what type of system and how much I/O are you doing.

You might want to look at statcpu to see how busy that is.

I.e do you want to serve data or managers?  :-)

[feedthemachine]

Thanks for your answer.

I am performing some sort of planning / POC and we are talking now "figuratively" (if you would).

I would like to know if there are any possible performance impacts during data collection. That's it. E.g. it would be great if I could have 2-3 probable scenarios when my performance can, in fact, be affected. I don't have a deep HPE storage knowledge yet, but judging from my previous experience some spikes can be seen when a good amount of performance data is collected.

Thus, I would highly appreciate any help with regard to the question above and some additional information of how could it be minimized (I know HPE is a great storage vendor, but I doubt when much data is collected the performance of the system remain untouched :) )

[Dennis Handly]

> I would like to know if there are any possible performance impacts during data collection

As I said it depends.  How much CPU and I/O load is on the array.

CPU is needed to move the data around.  The SR process are run on a different node than the master node to help.

But the I/O to the database still has to read/write to the .srdata VV.