Availability zones
Availability zones (AZs) are physically isolated locations (data centers) where cloud services operate.
There are multiple AZs within a region, each with independent power, cooling, and network infrastructure.
The choice of AZs is usually affected by the latency/proximity to customers, compliance, SLA, redundancy/data security requirements, and cost. All AZs in a region are interconnected for an easy resource replication and application partitioning.
Cross-availability-zone data distribution
Services can be replicated across multiple availability zones (AZs), which simplifies handling failures, decreases network latency, and enhances resource protection. This deployment model provides redundancy and failover in case an AZ goes down. Deploying services across multiple AZs enables smooth traffic transfer between AZs. It improves resiliency and reliability of workloads.
Aiven services across availability zones
For Aiven services, nodes automatically spread across multiple AZs. All Aiven's multi-node service plans are automatically spread among AZs of a region as long as the underlying cloud provider supports it.
The virtual machines (VMs) are distributed evenly across zones to provide the best possible service availability in cases an entire AZ (which may include one or more datacenters) goes down.
Cloud providers with AZs available to support your Aiven services are the following:
- Amazon Web Services
- Google Cloud Platform
- Microsoft Azure
- UpCloud
Supported availability zones
To learn what availability zones per cloud provider and region are supported for Aiven-managed services, check the Cloud column in List of available cloud regions.
With UpCloud, the only location where Aiven can automatically balance
replicas of services is upcloud-fi-hel
. For upcloud-fi-hel
, UpCloud
provides two datacenters (fi-hel1
and fi-hel2
). With a two-node
plan, for example, it will result in one of the servers in fi-hel1
and
the other in fi-hel2
.
Smart availability zones for Apache Kafka®
On top of spreading service's nodes across the availability zones of a cloud region, Aiven automatically balances replicas of your Apache Kafka® partitions into different AZs. Since Aiven automatically rebalances the data in your Apache Kafka® cluster, your data remains fully available when a node or a whole AZ is lost.