Avg response time Load runner Report vs Report from Apigee
I recently ran some tests in production env and observed a huge difference in avg response time report generated from load runner compared to report generated from Apigee. I’m using web Http protocol (rest API/XML calls). We have two diff Data center in diff locations. My load generators are in diff location and the endpoint I’m targeting is in diff location but I ran trace rt from load generators to other data center and latency is just 70 ms but response time diff is more than 10 sec. Test scenario -- protocol- web HTTP users—2k . Scenario-- Endpoints pointing to load runner in (datacenter A) – goes to ----- bigIp (datacenter B)-- Apigee (datacenter B)- ---- Target servers (datacenter B) and target server send back the response to apigee. And from Bigip to apigee its just 0.2 ms.
In apigee we can generate the report in which avg response under load is below 2 to 3 secs. But in load runner report avg response time is more than 10 sec. I’m trying to find out why there is a huge difference between the response time from apigee and load runner. Please let me know if there are any suggestions ...which I can try. Thanks in advance.
a few questions from to better understand your problem:
- Were the same high response times achieved when playing the script in vUGen?
- Did You minimize the runtime settings in the scenario?
a) Internet protocol - Preferences
- create snapshots
- request zlib headers
- Generate Web performance graphs (Hits per second and HTTP codes; Response bytes per seconds)
b) Browser emulation
- Simulate browser cache
- Download non-HTML resources
- Simulate a new user on each iteration
These settings can increase the transaction response time in LoadRunner
- How many Loadgenerators are serving the 2k vUsers?
- Did You monitor the system performance counters for Loadgenerators?
Did you notice any abnormalities (CPU, RAM, Network load)?