Scale an app and let it sleep

_{Updated October 19, 2026} You choose how many copies of your app run and how big each one is. An app can run a fixed number of instances, add instances as requests grow, or sleep when nobody uses it. Each running instance is billed at the app’s size.

Before you start

A website or API, a private service or a background worker. Scheduled tasks and one-off jobs have no scaling settings: see Run cron jobs and one-off jobs.
An app can run at most 6 instances, and your account 20 in total. Paused, failed and not yet deployed apps count toward the 20. Contact support to raise either limit.

Choose how it scales

Option	What it does	Available for
Fixed instances	Runs the same number of instances, 1 to 6, all the time	All three app types
Autoscaling	Adds instances between a minimum and a maximum as the requests in progress per instance grow	Websites and APIs, and private services that scale to zero
Scale to zero	Stops every instance when the app is idle and starts one on the next request	Websites and APIs, and private services

A background worker always runs a fixed number of instances.

Change the scaling

Open Settings

Open the app and go to Settings. Scaling shows how the app runs now.

Set instances or autoscaling

Fixed: leave Autoscaling off and set Instances.
Autoscaling: turn on Autoscaling, then set Min instances and Max instances.

TODO: Scaling section in app Settings with Autoscaling, Min instances, Max instances and Scale to zero

Let it sleep (optional)

Tick Scale to zero to let the app sleep when it is idle. On a private service, tick it first: Autoscaling then appears.

Apply

Click Apply scaling. A live app takes the change without downtime.

How sleeping works

The app stops its last instance about 15 minutes after its last request.
The next request wakes it. That first request takes a few seconds longer than usual.
On pay-as-you-go, a sleeping app costs nothing. The minutes it stays awake are billed like any running time.

Scale to zero suits test sites, internal tools and APIs with little traffic. Keep it off for an app that must answer at once, or that does work between requests. Websites, and private services that scale to zero, cannot listen on ports 8008, 8012, 8013, 8022, 9090 or 9091. The platform reserves them.

Change the size

Open Settings, Instance size, pick a size and click Change tier. The app restarts on the new size, and every instance uses it.

Size	vCPU	Memory	Disk	Per hour	Per month
Micro	0.25	512 MB	2 GB	$0.00411	$3.00
Starter	0.5	1 GB	5 GB	$0.009575	$6.99
Standard	1	2 GB	10 GB	$0.023274	$16.99
Pro	2	4 GB	20 GB	$0.038342	$27.99
Pro Max	4	8 GB	40 GB	$0.084918	$61.99
Scale	8	16 GB	80 GB	$0.16437	$119.99

Prices are per instance. The disk is wiped on every deploy. For an app on a monthly plan, size changes go through support.

How scaling is billed

Pay-as-you-go: every running instance is billed per second at the app’s size. With autoscaling, you pay for the instances that actually run.
Monthly plan: the plan covers the instances you bought at checkout. Instances above it, with a higher fixed count or autoscaling maximum, are billed as usage.

A spend cap emails you at 75%, 90% and 100% of a monthly limit, and can pause your apps when it is reached.

Do it from your tools

$APP_ID is the app’s ID from List app services or raff apps list. $RAFF_API_KEY is an API key. API: Scale app service autoscaling_enabled and scale_to_zero apply only when you send autoscaling_set: true. With it, leaving either one out turns it off.

# 2 fixed instances, always on
curl -X POST "https://api.rafftechnologies.com/api/v1/apps/services/$APP_ID/scale" \
  -H "X-API-Key: $RAFF_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"autoscaling_set": true, "autoscaling_enabled": false, "scale_to_zero": false, "replicas": 2}'

# Autoscale between 1 and 4 instances
curl -X POST "https://api.rafftechnologies.com/api/v1/apps/services/$APP_ID/scale" \
  -H "X-API-Key: $RAFF_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"autoscaling_set": true, "autoscaling_enabled": true, "min_replicas": 1, "max_replicas": 4}'

# Autoscale up to 3 instances and sleep when idle
curl -X POST "https://api.rafftechnologies.com/api/v1/apps/services/$APP_ID/scale" \
  -H "X-API-Key: $RAFF_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"autoscaling_set": true, "autoscaling_enabled": true, "min_replicas": 1, "max_replicas": 3, "scale_to_zero": true}'

autoscaling_target sets how many requests in progress one instance takes before another starts. To change the size, get the size IDs from List Apps tiers and send the new tier_id:

curl -X PATCH "https://api.rafftechnologies.com/api/v1/apps/services/$APP_ID" \
  -H "X-API-Key: $RAFF_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"tier_id": 3}'

CLI

# Fixed instances (for an app that does not autoscale)
raff apps scale $APP_ID --replicas 2

# Autoscale between 1 and 4 instances
raff apps scale $APP_ID --min-replicas 1 --max-replicas 4

# Autoscale up to 3 instances and sleep when idle
raff apps scale $APP_ID --min-replicas 1 --max-replicas 3 --scale-to-zero

# List sizes and their IDs
raff apps tiers

Products Home

Build

Store

Network

Manage

AI

Scale an app and let it sleep

Before you start

Choose how it scales

Change the scaling

How sleeping works

Change the size

How scaling is billed

Do it from your tools

Next steps

Set an Apps spend cap

View logs, metrics and a shell

​Before you start

​Choose how it scales

​Change the scaling

​How sleeping works

​Change the size

​How scaling is billed

​Do it from your tools

​Next steps

Set an Apps spend cap

View logs, metrics and a shell

Before you start

Choose how it scales

Change the scaling

How sleeping works

Change the size

How scaling is billed

Do it from your tools

Next steps