Feature Request to help answer the Q: "What proc. is using my disk!"

[hal900099]

When I'm using my computer in a very un-demanding way, I sometimes get unexplained slowdowns.  On certain occasions when this happens, I look at my hard drive light and see it's full on.  But, the problem is, I haven't asked my computer to do anything.

So the question is, *who* (what process) is using my computer? 

Right now, to try to answer this question, I currently look at Win XP's Task manager's "I/O Read Bytes" and "I/O Write Bytes" columns to see if I can find a number that is increasing in leaps and bounds.  This isn't a very good way to find the name of the offending program since all programs' values in these columns typically have non-zero (and often large) numbers in them totaled up from all th previous I/O done by the program.

However, I've noticed that the  "PF Delta" (Page Fault Delta) column usually has all zeros for all tasks -- except when a task has a burst of CPU activity and shows an easy-to-see non-zero value. 

To use this eye-catching non-zero aspect for disk I/O -- if one were to show the *difference* of bytes of I/O done between the last two time steps, most processes would show a zero, like a typical "Page Fault Delta". 

However, when a task was doing a lot of I/O it would consistently show an easily-seen non-zero I/O byte count and the unknown disk-I/O-using program could be discovered and any corrective action needed could be taken.


I suspect that this display of info would be much less difficult to implement than trying to eliminate I/O bottlenecks, as discussed in other threads...

Does this seem like a worthwhile feature to you?

- Hal

[Jeremy Collake]

Yes, it is a worthwhile feature, and may be implemented at some point. Unfortunately, Microsoft's new Resource and Performance monitor in Vista+ is the best thing since sliced cheese at easily showing you what is using any of your resources (specifically - CPU, Memory, Disk, Network). So, to some extent it has already been done .. at least for those who have upgraded from XP ;o.

[Jeremy Collake]

I may have missed your original intent.

I should also note as far as page faults (which their resource monitor also depicts), there are 'soft' and 'hard' ones that aren't as easily distinguishable. The hard ones actually incur a disk read, where-as the soft one's grab from some intermediary cache.

As for taking *action* on processes with excessive page faults ... Well, that's something I will research.. but I'm not sure it is 'the answer'. It really depends more on how much RAM is installed on your system, since a page fault is where virtual memory is accessed that is not already in RAM, inducing a page-in operation.