There are three, overpowering, reasons why load tests ought to be run against a creation site, as opposed to a clone of the site:
- Validity of the outcomes
- Cost regarding time and cash
- The fire-drill esteem
- Legitimacy of the Results
Most Web locales are carried out utilizing a wide assortment of programming, equipment, and administrations. The product ordinarily incorporates working frameworks, application workers, databases, and so on, the equipment incorporates firewalls, switches, load balancers, workers, and so on, and the administrations may incorporate substance circulation organizations, promotion workers, Mastercard confirmation frameworks, and so forth These parts come from various sellers, and every segment has an extraordinary presentation and adaptability profile.
To make matters considerably more convoluted, Rich Internet Applications utilizing Web 2.0 advancements, for example, AJAX and Silverlight present the customer side cycles, which may likewise be associated with accomplice applications. In this way, except if the framework under test is incredibly basic, making a clone of the creation framework with similar execution and versatility attributes as the first is exceptionally troublesome. Since a distinction in any of these parts can significantly change the adaptability of the whole framework, the consequences of any load test applied to the clone framework cannot be applied to the creation framework with any level of certainty.
- Cost in Terms of Time and Money
Regardless of whether the creation framework can be cloned, the expenses related with reproducing such a framework with the exactness needed to make the load test results on the clone be material to the creation framework will be critical. Since to be substantial, a total Web website load test should incorporate all segments (facilitating organization, firewall, switch, load balancer, and so forth) you would need to buy, introduce, and design every single segment. A few group might be enticed to clone a subset of their framework; rather than cloning the load balancer and four workers, for instance, they choose to load test only one worker and afterward increase the versatility result by four to show up at their adaptability of their whole framework.
This load balancing software would be a substantial test for the versatility of one worker, however not of the Web website all in all. This load test would not distinguish any issues with the load balancer, for instance, and would not affirm that the facilitating organization could deal with multiple times the load utilized in this more limited size test. Over long periods of business testing we have seen various load balancer issues (for the most part because of mis-setup) and various amazements when the transfer speed necessities at a facilitating organization were pushed.