h

IO

mon

™

  The File I/O Performance Monitor

Home  |  Products & Services  |  Contact  |  About hyperI/O^sm

Backgrounder
Big Picture
Product Info
Features
Architecture
Screen Shots
Documentation
Support
Download

Best Practices for using hIOmon

The hIOmon File I/O Performance Monitor software package provides an extensive array of features and functions, including a comprehensive set of file and device I/O operation performance metrics, a variety of interfaces, and numerous configuration options.

The following "best practices" represent guidelines, recommendations, and suggestions that can help you make effective use of the many capabilities offered by hIOmon.

Getting started with hIOmon

When starting out with hIOmon, it is important to get a basic sense of what hIOmon is about, including how it operates, what it can do, and what types of tasks it is suited for.

The following information can help you gain a basic understanding of hIOmon -- and without too much reading!:

• hIOmon Backgrounder (Why hIOmon?)
  

• hIOmon Big Picture (What is hIOmon?  What does it do?  What makes it unique?)
  

• Basic Key Storage Performance Questions (and how hIOmon can help provide answers) 

Quick Starts and Tutorials

The hIOmon software package is very comprehensive, with its support of several client interfaces, a robust and wide set of I/O operation performance metrics, a large range of features and functions, extensive documentation, etc.

So where to start, especially for beginners?  Read the documentation - oh no!

Most probably you would prefer a "quick-start" approach.  The "Getting Started" information provides several pointers along these lines, including several "Quick Starts" with "step-by-step" instructions to help you get started with hIOmon.

Also be sure to take a look at the short tutorials.  These tutorials can provide you with a sense how powerful hIOmon can be (and easy to use).

And do take heart that while the overall hIOmon package can perhaps be overwhelming to the newcomer, the various capabilities provided by hIOmon can come into play (and be found especially useful) for those folks that need (or eventually find that they need) to drill down so as to uncover some of the hidden mysteries of what exactly is going on with the specific file and disk I/O activity driven by their particular applications!

Top-Down Approach

When thinking about disk storage I/O performance, you'll most probably thought in terms of disk I/O operations down at the disk device (or perhaps disk subsystem) itself.  You likely haven't thought much about file I/O operation performance metrics, largely because such metrics were hard to come by, if available at all.

hIOmon changes the picture.  You can now collect, display, and export file I/O operation performance metrics upon a specific file and, moreover, summarized basis.  Furthermore, you can collect these empirical metrics using your very own applications and associated files just as you normally do -- with no benchmarking required!

So where to begin, especially since you probably have many thousands of files?  Won't trying to monitor all of those files give me a mountain of metrics, and maybe even cause my system to crash?

Of course, not all of those files are being used, especially at the same time.  Well how many are actually being used?  That's one question which hIOmon can help answer.

One approach is to configure the hIOmon software to collect summary metrics up to a certain maximum number of files and see if that number is ever reached.  You might be surprised at how many (or few) files are actually used.  And note that by using the "summary" metrics feature option uniquely provided by hIOmon, the amount of metrics collected (along with the overhead involved in collecting the metrics) is greatly reduced.

Other factors to consider beyond the number of files actually used include how the individual file themselves are actually being used and by whom (i.e., which processes/applications).  Here again, the summary metrics provided by hIOmon can help you easily and quickly establish a "characterization" of how particular files are being used and by whom, and with minimal overhead (including work on your part).

When thinking along the lines of application and file I/O activity characterization (and to frankly acknowledge the usual "crisis of the day" demands), it can be helpful to take a "pain points" or "Top Ten" approach to identify the primary applications and files that are of immediate concern.

hIOmon provides several features that can help you in this regard.  For example, the hIOmon "summary" metrics feature provides an option whereby the summary metrics are collected by hIOmon but only offloaded/exported if and when an "alert threshold" (e.g., when a total read I/O operation count, a write maximum response time, a total amount of read data transferred, etc., all of which you can specify) is detected by the hIOmon I/O Monitor component.  Accordingly, the hIOmon summary metrics available for display, exported, and further processing are thereby limited to those files satisfying a particular metric of interest to you  --  and again with significantly reduced overhead both upon your system and upon you.

The hIOmon "Top Ten" feature support also can be very helpful in this regard.  By using, for instance, the hIOmon WMI Browser, you can easily and quickly determine in real-time your particular "Top Ten" files, device, and processes based upon over 100 different sort-order criteria.

The overall key idea here is to take a "top-down" approach by leveraging the hIOmon "summary" metrics feature to first characterize your particular file and device I/O activity (and tied to the associated processes/applications), with a particular focus upon those specific metrics that are of particular interest to you (for example, response times, number of I/O operations and their rate (IOPS), the amount of data transferred, queue lengths, system file cache hit/miss and random/sequential access activities, etc.).

Back to top.

I/O traces  --  I don't need no I/O traces!

The hIOmon unique "summary" metrics feature affords you the luxury of having to deal with I/O operation trace data on but an usually rare occasion.  (By the way, hIOmon also provides an optional "I/O Trace Data" feature that you can use to collect a trace of individual I/O operations if you really need it).

Instead of struggling with reams and reams of I/O operation trace data, simply let hIOmon efficiently and automatically collect summary metrics so that you can, for instance, find out at a glance what the specific minimum, average, and maximum read I/O operation response times are for a particular file (or for a particular process/application or device overall).  The same goes for dozens of other metrics such as data transfer size, queue lengths, system file cache activity, random/sequential accesses, etc.

At the risk of repeating ourselves, the hIOmon "summary" metrics feature helps provide significantly reduced overhead both for your system and for you.  Save yourself (and your system) the time, effort, cost, and aggravation of dealing with volumes of I/O operation trace data.  Resist the urge to collect "all" of the individual I/O operation data and rely upon the hIOmon summary metrics to help you easily and quickly determine, assess, and evaluate the actual storage I/O performance of your files and their associated applications using the hIOmon "summary" metrics and a "top-down" approach.

Back to top.         

Where can I learn more about hIOmon?

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

• "Big Picture" information about hIOmon and the "Key Storage Performance Questions".

• 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) and tutorials.

• hIOmon benchmarking information.

Back to Top