Regions and zones

The evroc compute service uses a multi-zone architecture to provide high availability for your workloads. Zones are physically separate data centres within a region, each with independent power, cooling, and network connectivity. This separation protects your applications from localised failures.

How zones work

A region (such as se-sto for Stockholm) contains multiple zones. Each zone operates independently, so a failure in one zone doesn't affect resources in other zones. The Stockholm region has three zones:

  • se-sto-a
  • se-sto-b
  • se-sto-c

When you create certain resources, you specify which zone they should run in. This gives you control over where your workloads are deployed and how they are distributed for availability.

Regional and zonal resources

Resources in the evroc compute service fall into two categories based on their scope.

Zonal resources

These resources exist in a specific zone. When you create them, you choose which zone to use.

  • Virtual Machines - Run in a single zone
  • Disks - Stored in a single zone and can only attach to VMs in the same zone
  • Placement Groups - Distribute VMs across physical hosts within a single zone
  • Subnets - Each zone has its own default subnet

Regional resources

These resources span all zones in a region and aren't tied to a specific zone.

  • VPCs - Provide network connectivity across all zones
  • Public IPs - Can be attached to VMs in any zone
  • Security Groups - Apply to VMs regardless of which zone they are in

Cross-zone communication

VMs in different zones can communicate with each other over the private network within a VPC. Traffic between zones uses private IP addresses and is subject to security group rules, just like traffic within a single zone.

This cross-zone connectivity allows you to build multi-tier applications with components distributed across zones for resilience.

Designing for high availability

Distributing your workloads across multiple zones improves availability. If one zone experiences an outage, resources in other zones continue operating.

To take advantage of multi-zone availability:

  1. Deploy redundant instances of your application in different zones
  2. Use regional resources (Public IPs, Security Groups) that remain available even if a zone fails
  3. Run a load balancer to distribute traffic across instances in different zones

The evroc compute service doesn't include a managed load balancer. Deploy your own load balancing solution (such as HAProxy, NGINX, or a cloud-native option) to distribute traffic across VMs in multiple zones.

Next steps

  • Learn about VPCs and how they provide cross-zone networking
  • Understand Subnets and how each zone has its own subnet
  • Configure Security Groups to control traffic between zones
  • Use Placement Groups to spread VMs across physical hosts within a zone