All our multi node service plans are automatically spread among the availability zones of a region as long as the underlying cloud provider supports it (listed below). 

The virtual machines are distributed evenly across the zones in order to provide best possible availability guarantees in cases when an entire Availability Zone (which may include one or more datacenters) becomes unavailable.

Cloud Providers in Aiven with Availability Zone support:

  • Amazon Web Services
  • Google Cloud Platform
  • Microsoft Azure
  • UpCloud*

Smart availability zone placement for Kafka

In addition to having a service's nodes spread among the availability zones of the cloud region where its placed, we also automatically balance the replicas of your Kafka partitions into different availability zones. 

As we automatically rebalance the data in your Kafka cluster one of the many things we take into account is to ensure that if a node or even a whole availability zone is lost your data still remains fully available.

*UpCloud Availability Zone

In Upcloud's case, it only applies to upcloud-fi-hel which Upcloud provides the datacenters (fi-hel1 and fi-hel2). For example, with a two node plan, it will result with one of the servers in fi-hel1 and the other in fi-hel2.


Did this answer your question?