High and erratic server response time.

[Sumanta Bhowmik]

I am analyzing my site's landing page using pagespeed.The response time gets as high as 3 seconds on F4, but some times though rarely it is within the accepted limit. Instance is always up.

Checking at app engine dashboard I found that it does not show any cpu utilization on F4.The memory utilization is always below 100 MB.This means than using F4 is just overprovisioning. Peak traffic sent(cached) was 14kb/s while peak traffic sent was 4kb/s (at a different time.) 

I do not understand what is causing this high response time. Any insight or suggestions will be useful.

[Jordan (Cloud Platform Support)]

Response time is dependent on a lot of factors. Mainly your runtime, scaling settings, and your code. 

If you are using App Engine Flexible it will use Compute Engine virtual machines for its instances which take a bit longer to start compared to App Engine Standard. Your choice of runtime also affects the speed of response times as the language itself can slow down the process. 

Using auto scaling is also recommended as it will automatically scale up more instances when your traffic requires it. You can of course configure the max concurrent requests and max pending latency to optimize the amount of requests that are handled per instance, and change the amount of time a request is allowed to wait before forcing App Engine to start a new instance to handle it.

The largest factor to speed is of course your code. The amount of work you perform per request is the amount of time the request must wait. You can use the Cloud Trace tool to break down each request to better understand where all of the time was spent.  

Another major factor is networking. The location of your App Engine project compared to that of the clients making requests to your app can drastically change the time it takes for a response to be received.  

[Sumanta Bhowmik]

Thanks Jordan. Thats a good checklist.