How do you know the right value and are system interrupts delaying response times. Main problem is that system interrupts can delay performance with, for example, 10.000 interrupts, while on a other system you have 30.000 interrupts and no delay at all.
What are system interrupts. Think of this; you are working on writing a report, now the doorbell sounds and you need to open the door, you stop with your report and open the door. Next, you resume your report writing. Same is happening with your system, it stops working on one task and start working on another task.
Now we need to understand why your system has many interrupts; most of the times they are caused by drivers and of course, depending on the kind of drivers, causing this kind of interrupts, For example a invalid or out of date printer driver is a small problem. If the driver is your main disk driver, you have a bigger problem.
How can we avoid this kind of performance delay’s? Start with monitoring your system interrupts and create a base line of the average interrupts on every server. Now depending on your monitors and capability of setting KPI’s different for every server, you create a baseline of the system interrupts in a normal situation. Now you are able to monitor if this value suddenly goes up and if, for example, at the same time a driver update has occurred.
Also check your CPU load, if you have high CPU interrupts and high CPU at the same time and if this is happening many times you need to check your drivers.
Is this only for Windows? No, on Linux or Unix systems the same event can cause high system interrupts: Check the following site http://unix.stackexchange.com/questions/25203/how-can-i-know-which-irq-is-responsible-of-high-cpu-usage.
30 january 2016