Show disk i/o size as a histogram. uses linux perf_events.
bitesize [-h] [-b buckets] [seconds]
This can be used to characterize the distribution of block device (disk) I/O sizes. To study block device I/O in more detail, see iosnoop(8).
This uses multiple counting tracepoints with different filters, one for each histogram bucket. While this is summarized in-kernel, the use of multiple tracepoints does add addiitonal overhead, which is more evident if you add more buckets. In the future this functionality will be available in an efficient way in the kernel, and this tool can be rewritten.
Linux perf_events: add linux-tools-common, run "perf", then add any additional packages it requests. This also requires the block:block_rq_issue tracepoint, which should already be available in recent kernels.
Specify a list of bucket points for the histogram as a string (eg, "10 500 1000"). The histogram will include buckets for less-than the minimum, and greater-than-or-equal-to the maximum. If a single value is specified, two statistics only are gathered: for less-than and for greater-than-or-equal-to. The overhead is relative to the number of buckets, so only specifying a single value costs the lowest overhead.
Number of seconds to trace. If not specified, this runs until Ctrl-C.
Trace read() syscalls until Ctrl-C, and show histogram of requested size:
# bitesize syscalls:sys_enter_read count
Kbyte range of the histogram bucket.
Number of I/O that occurred in this range while tracing.
ASCII histogram representation of the I/O column.
While the counts are performed in-kernel, there is one tracepoint used per histogram bucket, so the overheads are higher than usual (relative to the number of buckets) than function counting using perf stat. The lowest overhead is when -b is used to specify one bucket only, bifurcating statistics.
This is from the perf-tools collection.
Also look under the examples directory for a text file containing example usage, output, and commentary for this tool.
Unstable - in development.