|
Overview
The statistics view gives numerical information that allows the user to determine in a bottleneck has occurred.
The same information can be viewed in graphical form in the Graph
Tab. The statistics view allows you to view data values for high and low
level objects, either for an entire business case, or for an individual image or back-end script. The statistics
view consists of a test results browser, on top, and a test detail table below.
The user interface for the test results browser uses a drill-down metaphor to navigate through the tree of performance
data. Clicking on an item in one of the top lists will display details about that item in the list to the right.
The meanings of the lists are given here:
Tests
A list of the performance tests that have been performed. Clicking on
a test will display the performance statistics about that test in the table below.
Business Cases
A performance test can be run against either a simple business case, or a load profile, which consists of more
than one case. This column contains a list of the business cases that were used in the test. Clicking on the name
of the business case will display the runtime statistics pertaining to that individual business case in the table
below. Note that the statistics gathered for a business case are only collected at the times that a business case
repeats. This means if a case hasn't repeated at all, there won't be an average time or any other statistics for
it, including errors.
Web Pages
A business case consists of a series of web pages. All of the web pages contained in the selected business case
are listed in the detail table.
URL
A web page can consists of calls to cgi-bin scripts and other back end processes, as well as images. All of the
low level parts of the selected web page will appear in this list.
The means of the individual statistics are given below. There are separate statistics
collected for the overall test. The individual statistics for business cases, web pages, and urls are the same.
Overall Test Statistics
When the user clicks on an item in the Test Column, the following statistics will be displayed:
| Time |
This column contains the time the values in the row were sampled. |
| Total Hits |
The total number of hits generated by all of the business cases at the current time.
Note that a hit is defined as being a request to the web site at the lowest level. A single web page will generate a
hit for each image the page contains as well as for each back end request the page contains. |
| Hits/s |
The average number of HTTP commands sent to the server every second at the current
time. |
| Bytes/s |
The average number of bytes that were transferred between WPTrainer and the web server
in each second during the sample period. |
| Cases/min |
The average number of business cases that were executed per minute during the sample
period. This gives some idea of how many transactions are being processed by your web site. For example, if one
of your transactions is the user making a purchase, this gives a rough idea of how many purchases per minute the
system can handle. |
| Users |
The number of virtual users that were active at this time in the performance test. |
| Errors |
The number of errors that were generated during this time period. Note that to view
the errors you have to select an individual URL, since that's the level where errors are generated by the web
server. (The web server only knows about individual URL requests, not web pages or business cases.) |
Statistics for Business Cases, Web Pages, and URLs
When the user clicks on a business case, web page, or URL, the following statistics will be displayed.
The leftmost column records the time at which the statistics were sampled. Each performance test has the option
of having a different sample period. Statistics are recorded when an event ended, not when it began. For example, when the
last byte is read from a GET for an image, the statistics for that event are recorded, keeping track of how long it took,
how many bytes were transferred, etc. When the sample period ends, all of the events are added up, averages are calculated,
and a statistic is added to the database. The same thing applies to higher level constructs such as web pages and business cases. This
means that you have to run a test long enough to complete at least one business case to have any statistics on it. This method
may give apparently sparse results if you are running very long tests with one user, but at normal test conditions which
have a numbers of virtual users,
there are business cases, web pages, and URL requests ending all of the time, which gives performance data in which
it is easy to spot performance problems.
Note that the format for the recorded performance statistics durations is MM:SS.MS, where MM is minutes, SS is in seconds, and
MS is in hundredths of a second.
| Time |
This column contains the time the values in the row were sampled. |
| Repeats |
The cumulative number of times the business case, web pages, or urls were repeated.
Each virtual user is assigned a business case to execute over and over again, and this number tells
how many times each business case, web page, and URL was repeated. Note that the term "virtual user" only means a separate thread of execution.
To the web server it would have appeared as if separate, unrelated users were repeating the test case. |
| Avg TTFB |
The average time to first byte, or the time it took for the web server to first respond
to the HTTP requests. |
| Min TTFB |
The minimum time to first byte during the time period. |
| Max TTFB |
The longest time the web server took to respond to the HTTP request during the time
period. |
| Avg Dur |
The average duration it took for the virtual user to receive all of the business
case, web page, or URL back from the web server. |
| Min Dur |
The minimum time it took for the virtual user to receive all of the business case,
web page, or URL back from the web server. This is the fastest a user would have viewed the results. |
| Max Dur |
The maximum time it took for the virtual user to receive all of the business case,
web page, or URL back from the web server. This is the slowest a user would have viewed the results. |
| Bytes/s |
The average number of bytes that were transferred between WPTrainer and the web server
in each second during the sample period. |
| Errors |
The number of errors that were generated during this time period by either a business
case, web page, or URL. Note that to view the errors you have to select an individual URL, since that's the level
at where errors are generated by the web server. (The web server only knows about individual URL requests, not
web pages or business cases.) |
Viewing Playback Errors
To view the errors that may have occurred during playback, select the Error Tab:
A list of errors generated by the entire test will be displayed as they occurred in time. If this table is viewed during a performance
test the errors will be updated automatically once a minute. Note that this display isn't affected by the
"drill down" selections, and instead is only affected by the test selection.
To get more information about an error, double-click the line in the Error Table. This will show a dialog that shows
details about the HTTP Request and Response (if this information is available). Below is an example of the HTTP Request
details ('Request' tab):
The top section shows the headers that were sent to the web server. The bottom section shows any content
that was sent (in this example, some form field names and data values are visible). This information
can help you determine the value of any parameters (URL parameters, form field, cookies, etc.) that were
sent to the server that could cause the error. Below is an example of the HTTP Response details ('Response'
tab):
The top section shows the headers returned by the web server (including the HTTP
response code). The bottom section shows the content returned by the server. Often,
the web server will return a web page that describes the error (and if you're lucky,
a remedy). This is the web page that a real user would have seen if they encountered
this problem.
The content shown can also be viewed in your browser by selecting the "View in Browser"
button. This will instruct the browser to load the error page from the test results and
allow inspection of the page that a user would have seen during the test. Note that only
the single page content is saved during the playback process - if the page references
URLs (for images, etc) that are only available during the context of a business case, they
may not be visible when viewing the error page after a test has completed. This depends
entirely on the design of the system under test.
|
