The average queue size is a good metric for how much a disk is working at any given time. It is a better metric then the Read- and Write rates, as those fail to take into account that reading 100MB sequentially is keeping a disk much less busy than reading 100MB of random 4k blocks. If a user wants to know how busy his (hard) disk currently is, the avereage queue size is the most meaningful metric. Therefore I propose that plasma-systemmonitor should offer a sensor for this metric for every physical disk. A possible source of data would be iostat (which labels is as aqu-sz)
I would like to second this. I've recently moved to Plasma from Windows and this is a very useful metric to have particularly for magnetic disks.