Simulating Irregular Response Latencies (Think Time)

One benefit of using virtual services when testing is to introduce irregular behavior that's difficult to get a real web service to produce on demand. A test script rushes through the test steps at the same speed every time, but in BlazeMeter, a transaction can also have a fixed or random-length delay before the response is returned.

BlazeMeter's Think Time option introduces a delay that can be used as a factor to simulate slow human responses or network latency. The delay can be a fixed value or can be sampled from a random distribution which makes it easy for you to simulate different types of downstream latencies.

There are several ways how to access the Think Time tab to add a delay to a transaction:

• Go to the Asset Catalog tab, open the transaction, and edit the Response Parameters.
• Go to the Service Virtualization tab, open a virtual service, and go to the Parameters tab.

For each transaction response, you have the following options:

Think Time

• No Delay / Don't use transaction delay
  Sets the Think Time to 0 milliseconds. This is the default.
• Fixed Delay
  Lets you define a fixed response delay in milliseconds.
• Random Uniform Delay
  A uniform distribution simulates a stable latency with a fixed amount of jitter.
  Example: To simulate a stable latency of 20ms +/- 5ms, use lower = 15 and upper = 25.
  This option takes two parameters:
  • lower - The lower bound of the range, inclusive.
  • upper - The upper bound of the range, inclusive.
• Random Lognormal Delay
  A lognormal distribution is a pretty good approximation of long-tailed latencies centered on the 50th percentile.
  This option takes two parameters:
  • median - The 50th percentile of latencies.
  • sigma - Standard deviation. The larger the value, the longer the tail.