w wokku
Get Started
~/docs
/
scaling

# Box Tiers

Five box sizes covering hobby apps through production. The number you can run is decided by your plan; the size by the workload.

Updated 2026-05-29 · Edit on GitHub ↗

A box is one Dokku container — what the rest of the industry calls a “dyno”. Wokku has five box sizes; the number you can run is decided by your plan, the size by the workload.

The five sizes

Tier RAM vCPU Disk Notes
Sleeping 256 MB shared, sleeps Idle apps. Wakes on first request (~2-3 s).
Small 512 MB shared 1 GB Side projects, dev / staging, hobby apps.
Medium 1 GB shared 2 GB Production Rails / Django / Laravel at low-to-medium traffic.
Large 2 GB 1.0 dedicated 4 GB Production workloads with steady traffic + a worker.
X-Large 4 GB 2.0 dedicated 8 GB Heavy production (image processing, ML inference, big DB clients). Team plan add-on.

Sleeping / Small / Medium run on shared vCPU — burst capacity depends on neighbour load. Large and X-Large get dedicated cores.

How many you get

Each plan ships with a fixed quota of each size. See the full grid in Plans & Pricing.

Plan Small Medium Large Sleeping
Free 3
Solo 3 3
Pro 3 2 3
Team 3 2 1 3

Team can buy extras: Rp 300.000/month per Extra Large, Rp 600.000/month per Extra X-Large. No metering — flat add-on price.

Why “sleeping” is special: a sleeping box stays out of the active-box quota and costs nothing. The app idles its container; the next request wakes it in 2-3 seconds. Good for staging environments, demo apps, and low-traffic side projects.

Picking a size

Rules of thumb that work for most apps:

Dev / staging → Small (512 MB). Plenty for bin/dev workloads.
Production Rails / Django / Laravel at <50 req/s → Small or Medium.
Production at 50–200 req/s → Medium.
Background worker (Sidekiq / Celery / RQ) handling 100+ jobs/min → Medium or Large depending on per-job memory.
Image / video processing, ML inference, heavy DB-side reports → Large or X-Large.

When in doubt, start one size smaller and watch metrics for a week. The dashboard surfaces p95 memory and CPU at 5-minute granularity — easy to spot a tier that’s chronically under or over.

Changing size

From the dashboard: open the app → Scaling → Change Tier. Apply.

Wokku rolls the container at the new size with no downtime: the proxy holds new connections for a couple of seconds while the new container starts; in-flight requests finish on the old container before it terminates.

Mid-cycle tier changes are prorated in the next invoice — you don’t get charged a full month for an upgrade you did on day 28.

Bundling vs splitting processes

The number of boxes you need depends on how you arrange your processes:

Bundle web + worker into one container → 1 box. How to.
Split into separate process types → N boxes, scales independently.

Pick bundled for small apps; pick split for production where you want web and worker to scale at different rates.

Self-hosted

Wokku Community Edition has no tier system. You manage resources directly on your servers — pick container memory limits in your Dokku setup as you like.

Was this page helpful?