RabbitMQ Best Practices for High Availability on Cloud

Question

I'm planning to deploy RabbitMQ on Kubernetes Engine Cluster. I see there are two kinds of location types i.e. 1. Region 2. Zone Could someone help me understand what kind of benefits I can think of respective to each location types? I believe having multi-zone set up could help enhancing the network throughout. While multi-region set up can ensure an undisputed service even if in case of regional failure events. Is this understanding correct? I'm looking at relevant justifications to choose a location type. Please help.

Answer 1

I'm planning to deploy RabbitMQ on Kubernetes Engine Cluster. I see there are two kinds of location types:

1. Region
2. Zone

Could someone help me understand what kind of benefits I can think of respective to each location types?

A zone (Availability Zone) is typically a Datacenter.

A region is multiple zones located in the same geographical region. When deploying a "cluster" to a region, you typically have a VPC (Virtual private cloud) network spanning over 3 datacenters and you spread your components to those zones/datacenters. The idea is that you should be fault tolerant to a failure of a whole _datacenter, while still have relatively low latency within your system.

While multi-region set up can ensure an undisputed service even if in case of regional failure events. Is this understanding correct? I'm looking at relevant justifications to choose a location type.

When using multiple regions, e.g. in different parts of the world, this is typically done to be near the customer, e.g. to provide lower latency. CDN services is distributed to multiple geographical locations for the same reason. When deploying a service to multiple regions, communications between regions is typically done with asynchronous protocols, e.g. message queues, since latency may be too large for synchronous communication.