3:  How do you characterize the I/O behavior of your particular files/devices/applications?

Understanding exactly how your specific applications are actually using their associated files and devices (i.e., an application's particular "I/O profile") is a key factor in troubleshooting issues, evaluating proposed improvements and the impact of implemented changes, comparing technology alternatives and benchmark testing, etc.

As examples, do you know your particular application's:

• Average read and write I/O operation response times, and similarly the maximum response times and when they occurred?

• Read and write I/O operation counts, and similarly the average and maximum I/O operation rates (IOPS) and when the maximum IOPS occurred?

• Read and write data transferred amounts, and similarly the transfer sizes (minimum, average, and maximum, along with when the maximum transfer size occurred)?

• Average and maximum read and write data transfer rates (MBs), and similarly when the maximum MBs occurred? 

• Average and maximum read and write I/O operation queue depths/lengths, and when the maximum queue depth/length occurred?

• Amount of I/O operation activity that was satisfied by the system file cache, and similarly the amount of I/O activity that required “physical device” I/O operations (within the operating system) to complete the associated file I/O operations?

• Amount of I/O activity that reflected either random or sequential access to the file (or device)?  And not only in terms of separate read and write I/O operation counts but also the amount of data transferred? 

hIOmon provides a robust set of file and device I/O operation performance metrics so that you can discover how your particular, individual files and devices are actually being accessed and used by your particular applications.

This comprehensive set of I/O operation performance metrics goes far beyond basic read and write I/O operation counts, response times, throughput, and data transfer sizes/amounts to uniquely include such metrics as queue depths/lengths, idle times, system file cache metrics (e.g., "hits" and "misses"), optional associated "physical device" I/O operation metrics (including "split I/O" operation information related to fragmented files) along with random/sequential access detection, plus much more!

Moreover, one key hIOmon feature is its unique ability to collect, display, and export the I/O operation performance metrics upon a summarized basis.  hIOmon is the only software tool currently available that collects summarized metrics such as random and sequential access I/O counts and data transfer amounts along with system file cache metrics (e.g., "hits" and "misses").   

Rather than having to labor over reams and reams of data representing metrics for each and every individual I/O operation (i.e., "I/O trace" data), hIOmon allows you to efficiently collect aggregate I/O operation metrics for only your particular files (and processes/applications) of interest.  These aggregate metrics can then be offloaded for display and/or export either periodically (user-specified), when the file is closed/inactive, or when a user-specified "Alert" threshold (e.g., a particular total write I/O operation count, a maximum read response time, total amount of write data transferred, etc.) is detected.

The optional hIOmon "Summarized Data" feature helps provide significantly reduced overhead both for your system and for you.  The aggregated metrics collected by hIOmon enable you to easily and quickly focus upon your specific items of interest and concern, without struggling over mountains of I/O trace data and/or having to use various tools to generate "summarized" metrics from such trace data.  But in the event that you do need to track down some particular I/O operation activity, hIOmon also provides an "I/O Trace Data" feature option as well.

Lastly, it is especially important to note that hIOmon can capture its robust set of I/O operation metrics while you use your very own actual applications and files as you normally do -- and within your own particular computer system environment.  Moreover, hIOmon does not require any application or file system modifications nor does it require operating system code changes!

See the short presentation entitled "Looking at an Application's I/O Behavior" for actual examples of empirical metrics collected by hIOmon for common, everyday applications. 

Back to top

4:  How do I correlate the I/O performance of files to my particular applications?

hIOmon provides file I/O operation performance metrics aggregated upon an individual process/application basis so that you can view overall I/O operation performance for a specific process/application.  hIOmon also enables you to not only easily tell which specific files (along with their respective file I/O operation performance metrics) are associated with a particular process/application, but also the particular processes/applications associated with a specific file.

By being able to quickly and easily correlate file usage with their respective processes/applications, you can determine, for instance, which particular processes/applications have the "hot" files.  Similarly, you can identify which particular files are contributing to the performance degradation of a specific process/application.   

Back to top   

5:  What is the actual impact of file fragmentation upon I/O performance?

The "Physical Device Extended Metrics" provided by hIOmon include several metrics related to physical device "split I/O" operations (which can occur, for example, when the data requested by a file I/O operation resides upon noncontiguous areas of the physical device so that multiple physical device I/O operations must be performed to satisfy the file I/O operation).  These new metrics can help provide, for instance, empirical data (moreover based upon actual file I/O operations) regarding the issue of file fragmentation by helping to show the actual performance impact of accessing fragmented files and of file fragmentation overall.

Back to top

Where can I learn more about hIOmon?

For additional information about hIOmon, please see the following links:

• "Big Picture" and "Backgrounder" information about hIOmon.

• Overall product information about hIOmon, including product description, solutions, uses, operating system support, availability and pricing.

• hIOmon documentation available for download (including datasheet, readMe document, overview presentation, licensing terms, user guide, and white paper), tutorials, and "best practices".

• hIOmon benchmarking information.

Back to Top