Response time x response time measured by caller

Platform Open Q&A
1

I wanna know the difference between the “Response time” and the “Response time measured by caller”.
There is a service called by another service at the same cluster, the latency is small at this cluster but the difference is high when I choose the P95 view. Is it commun?

Response time:

vitorsanders_0-1680533774074.png
vitorsanders_0-1680533774074.png1894×833 38.3 KB

Reponse time measured by caller:

vitorsanders_1-1680533813716.png
vitorsanders_1-1680533813716.png1888×856 40.4 KB

Thanks!

2

Hi,

This is a two type of metrics (Service metrics and Client-side metrics):

Client-side metrics are measured from the time a request leaves your application to the time the response is received by your application.

Server-side metrics are metrics as the name suggests directly on the server layer

Examples of metrics :

Service metrics:

  • Response time
  • Request count
  • Successful request count
  • Failed request count
  • Error count
  • Failure rate
  • HTTP 4xx count
  • HTTP 5xx count
  • CPU time
  • IO time
  • Wait time
  • Lock time
  • Number of calls to databases
  • Time spent in database calls
  • Number of calls to other services
  • Time spent in calls to other services

Client-side metrics

  • Response time measured by caller
  • Successful request count as seen by caller
  • Failed request count as seen by caller
  • Failure rate client
  • Number request attributes

Below you have an explanation of how this thesis looks at application testing.

Client-side a/b testing is best used when you are deploying an internal team of marketers to run experiments without the need for them to touch coding. Server-side a/b testing is best used when the scope of an experiment is bigger, for example, changes need to be made at the product level.

Radek

3

Hello Radek

this is a very intersting topic for me. I have two additional questions here. Is there any documentation where you have found this list of metrics seperated by server and client?

Second: Which prerequisite has to be fulfilled to get the ‘response time measured by caller’ out of the Dynatrace API? I can’t find this metric there. I think that this is not the same as the built in metrics for client-side response time and should not be mistaken.

I hope you can help me to get further details.

Thanks and best regards

Jan

4

Hi Jan,

Hi,

You can find the types of metrics that are available in Dynatrace here: Built-in classic metrics — Dynatrace Docs - I’ve highlighted the section for Response Time for you right away.

In general, these metrics are commonly used in application monitoring tools. You can find a lot of information not only in our documentation, but also on the Internet. Here you have a cool example for client-side: Client-side metrics overview  |  Bigtable  |  Google Cloud Documentation
or
https://www.indeed.com/career-advice/career-development/server-performance-metrics

Response time measured by caller is builtin:service.response.client.

Regards,

Radek

5

Hello Radek

thank you very much for your explanation.

Where do you have the information from? The Dynatrace documention doesn’t tell anything about the intention of the metrics (following your link above that was already familiar to me). These metric descriptons would be essential for us.

Same for the connection of Response time measured by caller and builtin:service.response.client. Where is this relationship documented?

6

Having these metrics described better in the documentation would be incredibly useful. I just went down this same rabbit hole a few months ago and it was not easy to find the answers that should just be in the documentation