What does 'Highest active time' for disk activity in Windows resource monitor mean?

I know what the disk io, disk queue length and other measures are, but what does 'Highest active time' mean?

Is it the amount of time it is busy handling requests, or something else?

When it is high, does it mean the CPU is busy doing some IO work, or is it just indicating that the disk is busy handling requests?

I'm trying to work out if 50% active time means that 50% of the time the disk is either seeking, reading or writing, rather than the kernel is spending 50% of it's time servicing IO requests.

Edit Another quick data point here. If you look at the difference between an SSD and a physical disk, the SSD has significantly less activity, so I guess this really means the amount of time the operating system is waiting for the disk to respond and returning data.

Nick R

Posted 2009-09-23T16:26:42.160

Reputation: 575

Answers

I’m pretty sure it is the same as the Perfmon counter "PhysicalDisk _Total %time". (It tracks this counter in lock-step.)

Perfmon gives this description of that counter:

% Disk Time is the percentage of elapsed time that the selected disk drive was busy servicing read or write requests.

Basically, it is the percentage of elapsed time that the physical drive(s) were busy. So Even though your MB/s usage may be low, the % busy time can be high if the disk is heavily fragmented, or is otherwise thrashing on lots of small operations.

On a server where you would try to match the right level of high-performing disks to the workload, you would consider the disk subsystem to be a potential bottleneck if the % active (busy) time plateau-ed above 80%. On a workstation if this number is high, it just tells you that your disk performance is maybe what is slowing down your work.

As to your last question: I'm quite certain it is the physical disk and not any kernel or process time. While waiting on the physical diskthe kernel blocks whatever processes is waiting on that disk operation, and uses the CPU for whatever else needs doing.

DanO

Posted 2009-09-23T16:26:42.160

Reputation: 2 494

@TomTom Perhaps you could add an answer to clarify. I understand your point in a RAID, but the metric would seem to at least correlate strongly with poor performance, similar to what's described here. Idling your taxi does waste cash (engery.gov ppt).

– ruffin – 2014-12-02T14:13:24.667

Not totally. The problem is that I can keep a Hard disc active without a lot of activity. So, 100% active does not necessarily mean "delivering me all the data it can". – TomTom – 2014-12-02T14:22:05.737

@TomTom and DanO : hello, I've question. In my case my TFS build server is not using RAID but a single hard drive. The disk highest active time when active is mostly above 80%, often hit 100%. Is this a bottleneck? Are upgrading my hard-drive (SSD or dedicated hard-drive per build agent) will help? I don't have physical access to the machine right now because I need to propose it first to my manager. Thank you. – hendryanw – 2015-03-31T09:39:36.283

Commone sense tells you that something hitting 100% utilization is bottlnecked there. Any experience wiill tell you builds are IO intensive. Get an SSD - you will be surprised. – TomTom – 2015-03-31T18:45:12.330

I'm getting figures of up to 13,000 on a currently-busy server!

I'm thinking I'll go with % idle time (currently 0.05) – Michael – 2015-11-12T04:47:57.370

I think that rather than being the _Total it is the MAX of the individual drive statistics. Of course this would be the same thing if there is only one disk. If you look under the 'Storage' section of the disk tab you will see the 'Active Time (%)' column the 'Highest Active Time' is the MAX of this column. – Martin Brown – 2012-01-24T13:12:52.010

As a note - it is a totally usless statistics. Because "busy" is not "overloaded". Especially on servers with multi disc RAID subsystems that are capable of significant throughput and with a ton of separate IO streams the disc subsystem is pretty much always having SOME command in processing, just not relly anything that keeps it busy. THat is like measuring a taxi utilization by the time the engine runs - but it may only run on a parking place to keep the heating going in winter. I always measure Seconds/IO, which is a GOOD indication - when it grows, the disc subsystem is overloaded. – TomTom – 2012-05-30T09:16:04.837

This shows the percentage of time that the disk is not idle and is actively serving requests. If a disk is constantly running at very, very high levels (say, more than 80%), it may point to a storage-related bottleneck. If your users are seeing performance issues, and you’re seeing 100% active times, you might need faster or more disks.

Salman

Posted 2009-09-23T16:26:42.160

Reputation: 31

From Mahalo:

The highest active time is the percentage of time spent performing IO (actually processing time spent at the IO level). It can get to 100% if Vista is defragmenting or doing some other IO heavy operation and very little else is going on.

regan

Posted 2009-09-23T16:26:42.160

Reputation: 712

Is this the amount of time kernel processes are waiting for IO to complete? So it is not necessarily the amount of time the disk is actually working. I thought that most IO was simply a matter of sending a command to the disk and waiting for an interrupt back when the transfer into or out of a memory buffer had finally completed. – Nick R – 2009-09-24T06:51:06.610