Get hands-on experience with 20+ free Google Cloud products and $300 in free credit for new customers.

External HTTP(S) Load Balancer

I am trying to understand google http load balancer. Following are the important components of an http load balancer.

1.Forwarding rule (the ip address of the load balancer)

2.Target HTTP proxies.

3.URL map.

4.Backend instances group.

    This documentation (https://cloud.google.com/load-balancing/docs/https) says that unlike a network load balancer (pass-through service), an http load balancer is proxy load balancer. All of the TCP connections are terminated at the proxies of load balancer and then new TCP connections are generated to backend instances based on URL map. For the proxies of the an load balancer, we need to define a proxy-only subnet (the ip range which will contain all the proxies) and the backends will receive traffic from these proxies (so we need to define a forwarding rule that allows http traffic from proxies to the bankends). Therefore it is important to create a proxy-only subnet in the region (by using --purpose=REGIONAL_MANAGED_PROXY).

I understand this documentation page (https://cloud.google.com/load-balancing/docs/https/setting-up-reg-ext-https-lb)

What I don't understand is that why are we not creating the proxy-only subnet in the 

1.Global External HTTP(S) Load Balancer (https://cloud.google.com/load-balancing/docs/https/setup-global-ext-https-compute)

2.Global External HTTP(S) Load Balancer (https://cloud.google.com/load-balancing/docs/https/ext-https-lb-simple)

Please explain it

Thank you so much for reading.

Solved Solved
0 4 3,203
2 ACCEPTED SOLUTIONS

Hi @usman317 ,

Here's why I think global external HTTP/S load balancers don't need a proxy subnet:

The “location” where the incoming connections are terminated is different. In the case of global external HTTP/S load balancers, the incoming connections are terminated at Google Front End proxies. The Google SRE workbook has a diagram that shows this. In contrast, the incoming connections to regional external HTTP/S load balancers are terminated at Envoy proxies that are located within a region, and so need a proxy subnet within your VPC.

I hope this helps.

View solution in original post

If you wish to learn and try hands on labs, I recommend you try cloudskillsboost https://www.cloudskillsboost.google/catalog?keywords=Load+Balancer&locale=&page=1 

 

View solution in original post

4 REPLIES 4

Hi @usman317 ,

Here's why I think global external HTTP/S load balancers don't need a proxy subnet:

The “location” where the incoming connections are terminated is different. In the case of global external HTTP/S load balancers, the incoming connections are terminated at Google Front End proxies. The Google SRE workbook has a diagram that shows this. In contrast, the incoming connections to regional external HTTP/S load balancers are terminated at Envoy proxies that are located within a region, and so need a proxy subnet within your VPC.

I hope this helps.

Yea I understand the concept of envoy proxy but I have to learn what is Google front end 

If you wish to learn and try hands on labs, I recommend you try cloudskillsboost https://www.cloudskillsboost.google/catalog?keywords=Load+Balancer&locale=&page=1 

 

Yes sir, I was doing the lab when this question came to my mind