Huge elapsed time while sending request to target

[k_prathip]

When we ran performance test we found a very huge elapsed time (target.sent.end.timestamp - target.sent.start.timestamp) close to 40 seconds in a few requests. What really happens in apigee in between calculating these times. This is in Apigee edge public cloud installation.

@Abhiram M. Hemanna @AbhiramManjulaHemannaNataraja @Suresh @Anil Sagar @ Google @Dino-at-Google

[adevegowda]

@Prathip,

The elapsed time (target.sent.end.timestamp -target.sent.start.timestamp) refers to the time taken for "Request sent to target server". You can find details about what happens during this phase in this community post.

You may see high elapsed time in this phase if the target server is slow in responding to API requests. Can you please check if there was any issue with target server or if it was responding slow to Apigee Edge during your performance test ?

If you still need further help, raise a ticket with Apigee Support.

[k_prathip]

Thanks for the response. Unfortunately the link did not work.

The target server is working fine and we are getting success responses, but I wanted to understand what all operations happen between the target.sent.start.timestamp and target.sent.end.timestamp. when would apigee exactly log the target.sent.end.timestamp. will it wait for any network level signals to log the time.

[adevegowda]

@Prathip,

Not sure why the link did not work.

In the "Request sent to target server" phase the following steps happen:

1. Message Processor to establish connection with the Target/Backend Server.

2. If it is a secure connection, then it includes the complete SSL Handshake process.

3. Message Processor to send the request payload to the Target/Backend Server.

We have seen some issues where the target server takes long time to respond back to Message Processor during connection/SSL handshake or during the data transfer.

It this is happening for every request, then you can check few more things:

1. Is the target server in the same region as the Message Proccessors ?

2. If the issue reoccurs again, you can take tcpdump and analyse what's causing high elapsed time while sending the request to target server ?

[trunganh890]

There is a different about response time (elapsed time) between postman vs trace tool.
I called a proxy via postman & saw the elapsed time greeter than one in trace tool.

Anyone know the reason !?

[#6]

Postman is a rest client. The time you are seeing in the trace tool is from the request reaching apigee till response sent back by apigee.

Postman includes extra time of request reaching apigee infrastructure Load balancer, network latency and also response time taken from apigee to the postman application.

So, you will see this greater in postman than apigee trace tool.

[trunganh890]

Thank you Priyadarshi !