- Documentation
- Services
- Scaling
Scaling
Guara Cloud offers two types of scaling for your services: horizontal (more instances) and vertical (more resources per instance).
Horizontal scaling: replicas
Replicas are additional instances of your service running in parallel. When you increase the number of replicas, traffic is automatically distributed among them, improving availability and throughput.
With multiple replicas, your service stays online even if one instance fails — the other replicas continue serving requests while the instance is replaced.
Vertical scaling: CPU and memory
You can also increase the amount of CPU and memory allocated to each instance of your service. This is useful for applications that need more processing power or consume more memory.
Configuring in the dashboard
To adjust your service’s scaling:
- Go to the service page in the dashboard
- Navigate to the Resources tab
- Under Pod size, keep Automatic or pick Choose a size and set CPU and memory
- Under Scaling, choose Fixed replicas or Autoscaling (minimum, maximum, and CPU and memory targets)
- Click Apply changes
Changes are applied automatically. If you have more than one replica, the update is performed gradually (rolling update) to avoid downtime.
You can also change sizing and autoscaling from the CLI:
# Choose a pod size and run two replicas
guara scale --cpu 500m --memory 1Gi --replicas 2 --project my-project --service my-api
# Enable autoscaling
guara scale --autoscaling on --project my-project --service my-api
# Disable autoscaling
guara scale --autoscaling off --project my-project --service my-api
Limits by plan
Scaling limits depend on your plan. CPU and RAM are an account pool shared by every service: you can split the total between them however you like, as long as each pod stays within the plan’s largest pod. See Resource Limits.
| Resource | Starter | Pro | Max | Business | Enterprise |
|---|---|---|---|---|---|
| Total vCPU (account pool) | 1 vCPU | 2 vCPU | 4 vCPU | 6 vCPU | Custom |
| Total RAM (account pool) | 2 GB | 5 GB | 12 GB | 25 GB | Custom |
| Largest pod (chosen size) | 1 vCPU / 2 GB | 2 vCPU / 5 GB | 4 vCPU / 12 GB | 6 vCPU / 25 GB | 4 vCPU / 8 GB |
| Automatic pod burst | 0.5 vCPU / 256 MB | 0.5 vCPU / 512 MB | 1 vCPU / 512 MB | 1 vCPU / 1 GB | 4 vCPU / 4 GB |
| Replicas per service | 10 | 10 | 10 | 10 | 10 |
During the free trial, your account uses the Starter limits.
Automatic memory growth
If a service keeps crashing because it runs out of memory, the platform raises its pod memory to the next size automatically, as long as your plan has a larger size and your account pool has room. The change is saved as the service’s chosen size, and you’re notified in the dashboard and by email. If there’s no larger size or no room in the pool, the size stays the same and you’re notified so you can act.
Graceful degradation
If you hit your plan’s limits, existing services keep running normally. However, new resource allocations will be rejected until you free up resources or upgrade your plan.
For example, if your services already reserve your plan’s entire CPU pool, you won’t be able to increase a pod size or add replicas. But your existing services continue running without interruption.