hIOmon Software and Services

Screenshots

hIOmon I/O Trender Display

 

The hIOmon "I/O Trender Display" is included within all versions of the hIOmon Disk I/O Ranger software, including for both the hIOmon Workload I/O Profiler (WIOP) and the hIOmon Workload I/O Profiler for Files (WIOPF).  Screenshots of the hIOmon Disk I/O Display, which presents the I/O metrics within a user-configurable "Bar Chart" graphical format, can be seen here.

The hIOmon I/O Trender Display can be used to present - within a time-series, stacked graphic/chart format - various hIOmon "Performance Threshold Range Metrics" that have been retrieved by the hIOmon Disk I/O Ranger.  These metrics are shown by the I/O Trender Display in accordance with the various configuration options that are provided by the I/O Trender Display.  The I/O Trender Display graphs/charts are displayed within the bottom frame of the hIOmon Disk I/O Ranger "Main Display".

The Performance Threshold Range Metrics are a set of automatically-summarized I/O operation performance metrics that are specifically focused upon the ranges of I/O operations per second (IOPS), megabytes-per-second (MB/s) data transfer rates, response times (RT), I/O operation queue depths/lengths (QD), and data transfer sizes/lengths (DTS) actually observed by the hIOmon I/O Monitor software component for individual devices, files, and processes.

The IOPS, MB/s, QD, and RT metrics are automatically included within the summarized I/O operation performance information that can be collected by the hIOmon I/O Monitor.  The DTS summarized metrics can be optionally collected.

Also note that these hIOmon metrics can be automatically-aggregated by the hIOmon I/O Monitor at the time when the individual I/O operations are observed in realtime.

 
Performance Threshold Range Metrics Display

The screenshot below illustrates the display of the read I/O operation IOPS Range Count metrics for a physical device (DR1).  The various Range Count metrics are displayed in a "stacked", time-series format using lines with steps.  The metrics can be displayed separately for the read I/O operations, the write I/O operations, and upon a combined read and write I/O operation basis.

Each IOPS Range Count value reflects the total number of one-second intervals that the hIOmon I/O Monitor actually observed a total number of monitored I/O operations within the corresponding range that were successfully performed.  Only those Range Counts that have a non-zero value are shown within the I/O Trender Display.

Each IOPS Range Count datapoint upon the graph represents the percentage of time (i.e., of the overall collection period, which is the same as the "Observation Period") for the respective IOPS Range Count.  This percentage is the IOPS Range Count value divided by the total number of seconds that comprise the overall collection period.  As such, it basically indicates the percentage of time within the overall collection period that the respective IOPS Range Count value (e.g., "IOPS 1 - 99") was observed by the hIOmon I/O Monitor software component.  The hIOmon I/O Trender Display "tool-tip feature" can be used to display for a particular data point the specific percentage value for that IOPS Range along with the running total percentage (in brackets) for all of the IOP Range Count values.

 
hIOmon I/O Trender Display for Read IOPS

"Sub-Interval" headings can appear below the monitored item name for the top-most graph displayed.  These headings provide several hIOmon "DataTransferred/Time Index (DXTI)" values for the associated time interval report.  The length of the respective time interval can be displayed by using the tool-tip feature (i.e., hover the mouse pointer near the top of the heading).  The DXTI values shown reflect the lowest (blue), average (yellow), and highest (red) DXTI observed during the respective time interval. The left and right borders can line up with the starting and ending times respectively of the associated time interval (but please note that this is not always the case since space limitations can limit the size of these headings).

The solid blue bar below the "sub-Interval headings" reflects the overall collection period.  As shown within the screenshot above, the tool-tip feature can be used to display the overall collection period starting and ending timestamps along with the respective overall period length and average DXTI value observed.  Please note that due to space limitations, a single blue line can reflect more than one overall collections period; this generally occurs when two or more overall collection periods basically abut each other.

 

The hIOmon "I/O Trender Display" complements the "Disk I/O Ranger Display".  Its use of the time-series, stacked graphic format and symbolic color-coded ranges enable you to easily and quickly glean key "I/O Profile" insights at a glance.  Insights such as the overall amount of I/O operation activity, including the prevalence of low and high IOPS ranges, low and high response time ranges, etc.

 

Data Transfer Size (DTS) Summary Metrics

As shown by the screenshot below, the hIOmon I/O Trender Display also supports the display of the hIOmon "Data Transfer Size (DTS)" summary I/O operation performance metrics that can optionally be collected by the hIOmon software.  Note that these DTS metrics can be summarized in realtime by the hIOmon I/O Monitor.

The DTS summary metrics include, for example, a count reflecting the total number of times that the hIOmon I/O Monitor observed a read I/O operation whose data transfer size/length was that of the corresponding size/length.

The hIOmon I/O Trender Display uses a color-coded scheme to distinguish between the fixed, default set of data transfer sizes (i.e., the commonly-referenced data sizes starting with 512 bytes, 1 KiB, 2 KiB, 4 KiB, and continued doubling up to 8 MiB) and any user-specified data transfer sizes (e.g., 1536 bytes, 2560 bytes, etc.) for which summary I/O metrics can be collected by the hIOmon software.  See the Disk I/O Ranger help information in Documentation for additional details about these various data transfer sizes.

Each DTS Range Count datapoint upon the graph represents the respective DTS Range Count value divided by the total number of I/O operations of the respective type that successfully transferred data.  As such, it basically indicates the percentage of applicable I/O operations that used the respective data transfer size as observed by the hIOmon I/O Monitor.  The tool-tip for a particular data point displays the specific percentage value for that DTS Range Count along with the running total percentage for the DTS Range Count values.  The "running total" percentage reflects the percentage of applicable I/O operations that used the respective DTS or a DTS Range Count (i.e., a data transfer size) shown below it.

This screenshot also illustrates the variety of summary I/O metrics (e.g., I/O operation count; number of "fast" I/O operations, i.e., those whose response time was observed to be under one millisecond; maximum IOPS and MB/s; minimum, average and maximum response times observed; etc.) that can be displayed by the tool-tip feature and all of which are specific to the respective data transfer size.

The DTS display support provided by the I/O Trender Display additionally includes support for the extended configuration options (e.g., the "Transfer Amount", read/write I/O percentage, and the random I/O percentages).  Please see the I/O Trender help information in Documentation for additional details about these display configuration options.

 
hIOmon I/O Trender Display for DTS

 

The hIOmon "I/O Trender Display" can visually reveal quickly and easily the exact extent to which data transfer sizes such as 4K, 8K, 16K, 64K, 128K, etc. are actually used by your specific applications and with your particular devices during normal, everyday usage.

See how these empirical metrics line up with benchmarking reviews, vendor performance claims, and various "rules of thumb".  In addition, use the "tool-tip" feature to find out a variety of performance metrics (such as IOPS, MB/s, and response times) observed for a particular data transfer size.

 

Multiple Graphs

Two (or three) different graphs can be displayed concurrently as illustrated by the screenshot below where the write I/O operation IOPS Range Count values are displayed along with the write MB/s Range Count values and the Response Time (RT) Range Count values.

Each MBs Range Count reflects the total number of times that the hIOmon I/O Monitor observed an actual megabytes-per-second (MB/s) transfer amount within the corresponding range for I/O operations that were successfully performed.  For example, the write "MBs < 1" Range Count metric reflects the accumulated count of one-second time intervals during which the hIOmon I/O Monitor observed an actual total number of data bytes transferred by write I/O operations of one byte or more, but less than 1 000 000 bytes (i.e., 1 megabyte).  Only those MBs Range Counts that have a non-zero value are displayed.

Each RT Range Count reflects the total number of times that the hIOmon I/O Monitor observed an I/O operation whose response time (i.e., the time duration between the start of an I/O operation and the completion of the I/O operation, which includes both the service time and the queue time) was within the corresponding range.  Please note that only those I/O operations that were successfully performed are included within the RT Range Counts.

Each RT Range Count datapoint upon the graph represents the respective RT Range Count value divided by the combined sum of the RT Range Count values.  As such, it basically indicates the percentage of applicable I/O operations that experienced the respective Response Time ranges as observed by the hIOmon I/O Monitor.  The tool-tip can be used to display for a particular datapoint the specific percentage value for that RT Range along with the running total percentage (in brackets) including it and the RT Range Count values shown below it.  The "running total" percentage reflects the percentage of applicable I/O operations that experienced a response time within the respective Response Time range or below (e.g., for the third Write RT Range Count whose range is "100 us < 200 us", the percentage of the total number of write I/O operations whose response time was less than 200 microseconds).

 

hIOmon I/O Trender Multiple Graphs

 

The ability to display multiple graphs within the same I/O Trender Display enables you to more easily investigate possible correlations between the different I/O operation performance metrics (e.g., between IOPS and Response Times).

 

I/O Trender Display Options

The "Customize" button at the top right of the hIOmon I/O Trender Display can be used to display a "Settings Configuration" display panel/window (screenshot shown below) that allows you to limit, for example, the display to show only combined "Read and Write" I/O operation IOPS Range metrics.  It can also be used to limit the display to a selected Range Count type (e.g., only IOPS Range Counts).

hIOmon I/O Trender Display Settings

The hIOmon "I/O Trender Display Settings" window provides a variety of configuration options for displaying the hIOmon "Performance Threshold Range Metrics", namely:

  • "IOPS" for the ranges of I/O operations-per-second
  • "MB/s" for the ranges of megabytes-per-second data transfer rates
  • "Response Time" for the ranges of response times (RT)
  • "Queue Depth (QD)" for the ranges of queue depth/length

In addition, configurations options are available for displaying the hIOmon "Data Transfer Size (DTS)" metrics, which reflect a variety of summarized I/O operation performance metrics upon an individual data transfer size basis.

The hIOmon "I/O Trender Display" also supports a variety of configuration options that can be used to optionally include additional metrics within the displayed graph(s):

  • The "I/O Count" option, which enables the additional display of the overall total number of applicable I/O operations observed during the overall collection period for the respective file, device, or process.  Respective increments of the I/O operation counts are shown upon the right-side y-axis of the graph.
  • The "Transfer Amount" option, which is used to additionally display the total amount of data transferred by the applicable I/O operations during the overall collection period for the respective file, device, or process.  Respective increments of the transfer amounts are shown upon the right-side y-axis of the graph.
  • The "RW %" option, which is used to additionally display the read I/O operation percentage (for read and read/write displays) and the write I/O operation percentage (for write displays)
  • The "Random %" option, which is used to additionally display the "random access I/O operation percentage" and the "random data transferred percentage"
  • The "DXTI" option, which is used to additionally display the hIOmon "DataTransferred/Time Index (DXTI)" values
  • The "Active %" option, which is used to additionally display the percentage of time during the Observation Period that I/O operations were active for the associated monitored item (i.e., issued I/O operations being processed, including those queued)
  • A "Show interval deltas" option, which is used to display the metrics upon a per-interval basis rather than upon the total, cumulative basis, which is the default manner of display.  When this option is used, each data point reflects the change (e.g., increase or decrease) of the respective metric since the end of the preceding interval.

The screenshot below illustrates the use of several of these additional configuration options.

 

The "Synopsis" configuration option provides for the display of a brief, general summary of key performance values along with the associated time context.  The time context includes the respective "Time Scope", which is the overall time period associated with the performance values; the Time Scope can be a day, week, month, one or more quarters, or a year.

Three percentages are provided for each displayed Time Scope:

  • The "Observation Period" as a percentage of the Time Scope.  The Observation Period is the overall length of time during which the hIOmon I/O Monitor was running and capturing the displayed performance metrics.
  • The "Time IOPS" as a percentage of the Time Scope.  This percentage reflects the amount of time (based upon seconds within the Observation Period) during which I/O operation activity of the respective type was observed by the hIOmon I/O Monitor.  These seconds within the Observation Period reflect the accumulated count of one-second time intervals for all IOP Ranges combined.
  • The "Active Time" percentage, which reflects the percentage of the Time Scope during which the I/O operations were active for the monitored item (i.e., issued I/O operations being processed, including those queued).

In addition to the Time Scope and related metrics, the following I/O operation performance values are displayed for the associated Time Scope:

  • The "I/O Count", which is the total number of I/O operations of the applicable type observed by the hIOmon I/O Monitor for the monitored item during the Time Scope.
  • The "Data Transfer" amount, which is the total amount of data transferred by the I/O operations of the applicable type as observed by the hIOmon I/O Monitor for the monitored item for the monitored item during the Time Scope.
  • The overall hIOmon "DataTransferred/Time Index (DXTI)" value for the monitored item. The displayed DXTI value reflects the DXTI for the Time Scope.

Optionally, the following hIOmon "Performance Threshold Range Count" related performance values along with read/write and random access percentages can also be displayed by enabling the "Range Count" option:

  • For the "IOPS Range Counts", the IOPS Range Count range with the largest accumulated count.  In addition, the "highest" IOPS Range Count range is also displayed, i.e., the highest IOPS Range Count range with a non-zero accumulated count.  Note that the percentages displayed reflect the percentage of time within the Observation Period that the respective IOPS Range Count range was observed by the hIOmon I/O Monitor.
  • For the "MBS Range Counts", the MBS Range Count range with the largest accumulated count.  In addition, the "highest" MBS Range Count range is also displayed, i.e., the highest MBS Range Count range with a non-zero accumulated count.  Note that the percentages displayed reflect the percentage of time within the Observation Period that the respective MBS Range Count range was observed by the hIOmon I/O Monitor.
  • For the "Queue Depth (QD) Range Counts", the QD Range Count range with the largest accumulated count.  In addition, the "highest" QD Range Count range is also displayed (i.e., the highest QD Range Count range with a non-zero accumulated count) along with the "QD > 1" Range Count.  Note that the percentages displayed reflect the percentage of I/O operations of the applicable type that experienced the respective Queue Depth range as observed by the hIOmon I/O Monitor.
  • For the "Response Time (RT) Range Counts", the RT Range Count range with the largest accumulated count.  In addition, the "highest" RT Range Count range is also displayed (i.e., the highest RT Range Count range with a non-zero accumulated count) along with the "lowest" RT Range Count range.  Note that the percentages displayed reflect the percentage of I/O operations of the applicable type that experienced the respective RT range as observed by the hIOmon I/O Monitor.
  • "R/W %", which includes the read I/O operation percentage (for read and read/write displays) or the write I/O operation percentage (for write displays).  In addition, the read data tranferred amount percentage (for read and read/write displays) or the write data tranferred amount percentage (for write displays) is also displayed.
  • "Random %", which includes the read random access I/O operation percentage (for read and read/write displays) or the write random access I/O operation percentage (for write displays).  In addition, the read random access "data transferred" percentage (for read and read/write displays) or the write "data transferred" percentage (for write displays) is also displayed.

Please note that the Synopsis option can be selected separately for read, write, and/or combined read and write I/O operations.

See the hIOmon I/O Trender "Synopsis" Display screenshot for an illustration of the Synopsis display.

 

Please see the I/O Trender help information in Documentation for additional details about these display configuration options.

 

Key hallmarks of the hIOmon I/O Trender Display include its numerous display options centered around the hIOmon Performance Threshold Range Metrics and its flexibility to dynamically generate charts of particular interest to you.

Whether it be for real-time or historical displays, the I/O Trender Display provides a simple, quick, and easy way to help empirically uncover the specific "I/O Profiles" associated with your particular applications, workloads, and devices.