Appengine Dashboard: Instances (created, active, billed)

[Nilson Pontello]

I am trying to understand the difference between created, active and billed estimate from attached graph.

This project has exactly 17 microservices and before may 23rd they were all manual scaled instances, one for each service.

By 23rd I've enabled auto scaling for them all and it started counting "Billed Instance Estimate" but manual ones were also billed before although graph doesn't show them.

The number of created instances got high on the 23rd because I've deployed the same services several times.

My questions are:

1. Why manual scaled instance don't seem to appear here?
   
2. What is considered an active instance?
   
3. After enabling auto scaling, "billed estimate" jumped to 12, "active" struck as 1 and "created " is around 16. How those 3 numbers related to each other?

Thanks

[Jordan (Cloud Platform Support)]

1. The reason for manually scaled instances not having a Billed Instance Estimate line in the App Engine Dashboard 'instances' graph is due to the fact that you always pay for the amount of manually scaled instances you have set in your app.yaml (aka the blue Created line). Therefore, instead of having two lines always equal to each other, it is cleaner to simply display one line. 

2. The Active instances line in the graph represents the amount of instances that are currently using resources to respond to your incoming requests. Where as the Created line shows the amount of instances that are waiting to do work. 

3. Unlike Manually scaling, Automatic scaling will start and stop instances for you based on your application requirements. It will therefore not charge you for Created instances that have been idle for 15 minutes. You will also not be charged for any Created instances exceeding the amount of 'max_idle_instances' that you set for each service. Therefore, the Billed Instance Estimate line will appear for Automatic scaling in order to show you an estimate of what you are actually being charged (note that it is just a rough estimate based on averages and your actual bill will reflect the correct numbers).  

[Nilson Pontello]

Thanks Jordan but I am still a bit confused with the relation between Active and Billed Estimate. Please check if below affirmatives are correct:

• Instance will be counted as Active if it was processing at least one user request on the exact time statistic was collected.
  
• Instance will be counted as Billed Estimate since it became Active till fifteen minutes after the last request has finished processing. 
• Then GAE can keep it as Created (but Inactive) or just shut it down. I won't pay for such instances
• Any Created (but Inactive) instances can become Active to process the incoming request then it will start counting as above Billed Estimate

Regards

[Jordan (Cloud Platform Support)]

You are correct. The Billed Instance Estimate line will also always contain the amount of 'min_idle_instances' that you have configured in your app.yaml (much like manual scaling in that you tell App Engine you always want this minimum amount of idle Created instances waiting to accept requests).

[Petr Čada]

Hi guys,

I'm really glad that this discussion exists because I'm also quite confused with values which I see in our Google Cloud Platform console. Please look at the picture (Instances.png) which I sent in the attachment.

I understand that "Billed Instance Estimate" can be bigger number than "Active" (due to 15 min idle time) but what I don't understand is how "Active" can be bigger than "Billed Instance Estimate". In the picture which I sent in the attachment are values "Active" 24.0 and "Billed Instance Estimate" 13.0. This behaviour is quite common in our projects.

Can you please help me to explain how it's work?

Thanks a lot for help

Petr