Tests
|
|
|
|
How to measure URL load time How to measure URL load timeMeasuring Web site availability and performance is far more difficult than is generally realized. You can find a description of various approaches and what we do and why here. Test IntervalUnder normal circumstances we test your URLs once every 30 minutes. In our experience this provides good data on availability and performance as experienced by a user. We are occasionally asked if we will test more frequently than once every 30 minutes. The answer is a qualified yes. We subscribe to "good netizen" guidelines. These guidelines discourage web site monitoring that becomes offensive by imposing inordinate load on the target server without the permission of the site owner and site host. As a general rule, sampling every 30 minutes is permissible, while more frequent sampling is problematic. If you want us to sample your site more frequently we will do so with your permission and, if your site is hosted, the site hoster. Note, that while it is technically possible to test at almost any frequency, sampling more often than every few minutes is not generally useful. If you wish to monitor your URLs more frequently than every 30 minutes, please contact us and we can discuss what is involved and the various options available to meet your needs. Note that the failure and slow testing policy affects the testing interval. See here. AlertsMonitor helps you detect and respond to site problems more quickly, improving your customer's Internet experience. Monitor watches the URLs you specify and when there is a problem sends an email to one or more addresses. Failure AlertsA failure alert is issued when the site responds with an HTTP error such as 404, file not found, or when the site fails to respond completely within 30 seconds. Failures are noted and failure analysis included in your monthly reports. We are occasionally asked if we can change the failure timeout time. The answer is a qualified yes. We use 30 seconds since it represents the absolute upper limit of what most reasonable users will willingly tolerate and continue to use the site. If you have concerns in this area contact us and we will review the available options. Slow AlertsA slow alert is issued when the site responds slowly. The slow response time is set at 6 seconds. Slow responses are noted and slow response analysis included in your monthly reports. We are occasionally asked if we can change the slow time. The answer is a qualified yes. We use 6 seconds since it represents the upper limit of what most reasonable users will generally accept and continue to use the site. If you have concerns in this area contact us and we will review the available options. Single test protocolMonitor1 uses the single test protocol. It works like this
For most uses this protocol is entirely adequate. Immediate retest protocolIn many cases, a page that fails to load or loads slowly will be OK if you try again. Immediate retesting recognizes this and will retest the page immediately before sending an alert to you. This generally reduces the number of alerts you receive. Monitor2, Monitor3, and Monitor4 use the immediate retest protocol. An occasional retest is not a matter of concern. However, frequent retests are an indication of potential difficulties. Retesting events are reported on your reports. Delayed retest protocolSometimes the Internet will misbehave somewhere between the user's workstation and the institution's server. Oftentimes if you wait a few minutes and then retest, the problem will have resolved itself. Delayed retest notifies you of the test outcome, including if it is successful. Monitor3, and Monitor4 use the delayed retest protocol. An occasional retest is not a matter of concern. However, frequent retests are an indication of potential difficulties. Retesting events are reported on your reports. Resolution testingOnce a problem occurs, it is important to know when it is corrected. With resolution testing, if the delayed retest fails, we will repeatedly reschedule a delayed retest. You will continue to receive alert emails until the problem is resolved.Resolution testing is a no cost option with our Monitor4 service. Resolution testing results are included on your reports. |
|
Copyright 2004 Bell Curve Group. All rights reserved. Reproduction of content in any form without prior written permission is forbidden. Please read the terms of use for Bell Curve Group materials. |
