|
Administrator
|
This post was updated on .
You can use your own values for testing IOs.
can you predict how many IOs per sec you application will create?
If you know or if you can guess, then you can take those values as reference.
If you don't just start small value and increase it in every run.
Note the results and present them to your supervisor. You can say that for ex: The system performs well till it reaches XX number of IOs per sec.. (or something like that)
As for the RAID, you can check it with your Storage admin and OS admin. There can be a hardware level raid or a software level raid.. Software raid can be seen in OS layer, an the hardware layer (storage level or local physical disk level) can also be seen from OS by using some tools, but I suggest you to take help from your OS and Storage admin.. It is a long story for explaining here.
I use dd and AWR most of the time, but the tool named IO meter seems quite good, as well.
Read the documentation for IO meter to understand it properly and then start using it..
There are packaged tests in that tool. So you can just choose one like '8K random DB read' and test it..
First, learn your own env I/O characteristic and then use that tool accordingly.
I mean if your tablespace will be 8k and if you ll do tablescans, then you should test 8K random read..
ıf you will have index scans, then you should test 8K sequential read..
As for access specification, again choose a value accordingly. (you need to consider your OS and Disc IO sizes)
Anyways, I actually suggest you to run the Pre-loaded tests. They will hold their own Access patern and as well their own Test Setup.
Read the MOS document carefully and again, read the documentation of those tool seperately before using them.
Got the idea right?
|
Locked
