search cancel

Portal 4.4 documentation doubt

book

Article ID: 187999

calendar_today

Updated On:

Products

CA API Developer Portal

Issue/Introduction

In last paragraph: https://techdocs.broadcom.com/content/broadcom/techdocs/us/en/ca-enterprise-software/layer7-api-management/api-developer-portal/4-4/install-configure-and-upgrade/install-portal-on-docker-swarm/prepare-your-environment/deployment-topology.html

What exactly means: In a three-node deployment topology, the one Manager node runs only the dispatcher service. The two worker nodes runs the rest of the services for 

API Portal

. This scenario increases processing capabilities because the services are load balanced by your global load balancer or by DNS round-robin at the DNS server. When traffic reaches the Docker Swarm Manager node, further service load balancing is achieved by an internal application load balancer.

I get confused with "services are load balanced by your global load balancer or by DNS round-robin at the DNS server" and "further service load balancing is achieved by an internal application load balancer". Could you explain the exact behaviour, please?

Environment

Release : 4.4

Component : API PORTAL

Resolution

In any docker based system, scaling is partially implemented with routing rules that are created via a proxy component that also acts as a load balancer. This is configured with the rule sets that we supply.

However when you use multi-environment deployment, there needs to be something that makes it so the proxy in each environment can be accessed from the public internet.

That's implemented via a global load balancer or DNS.
In a GLB, there's a component that attempts to establish which server is 'closer' to the user and directs the network connection there.
In a DNS based system, a random entry is chosen, with failover should the first selected DNS address prove to be unavailable.