Dawnrose
Log in Start free Menu

Product Scale

Grow without rebuilding anything.

Give a service more CPU, more memory or more replicas. Dawnrose tells you when it needs them.

Start free Connect your agent
apiproduction3 replicas, 4 vCPU and 4 GB each

CPU, across replicas

100%50%030 min agonow

Replicas

  • replica 062% CPU
  • replica 158% CPU
  • replica 241% CPU

How far each plan goes, per service

  • Trial1 vCPU, 1 GB, 1 replica
  • Hobby8 vCPU, 8 GB, 3 replicas
  • Pro32 vCPU, 32 GB, 20 replicas

api needs more CPUOn Hobby

CPU has been at 96% of its limit (4 vCPU × 3 replicas) for over 10 minutes. Raise its CPU limit to 8 vCPU per replica (now 4 vCPU).

Sample data

One service, from quiet to busy.

CPU, memory and replicas are settings, not a rebuild: change one and it goes out with the next deploy. This is how far a single service goes on each plan, and where api is today.

apiWhat one service can have, plan by planSample data

CPU4 vCPU

Memory4 GB

Replicas3

api is at Hobby’s 3 replicas. A fourth replica takes Pro. On Hobby it can still grow to 8 vCPU and 8 GB.

Room to grow, with limits you set.

  • Replicas

    Up to 20 copies of a service on Pro and 3 on Hobby, behind one address, with requests shared between them.

    • replica 062% CPU
    • replica 158% CPU
    • replica 241% CPU

  • Advice from real usage

    When a service stays above 80% of its CPU limit or 85% of its memory limit for 10 minutes, you and your agent get a suggestion with the numbers behind it. It changes nothing by itself.

    web is running out of memory. Memory has been at 94% of its 1 GB limit for over 10 minutes. Raise its memory limit to 2 GB per replica (now 1 GB).

  • Spend limits

    Usage stops at the limit you set, so scaling up never surprises your bill. Reserving a lot of memory also needs a balance that covers it for 7 days.

    $7.40 of your $50 limit this period
    • Warnings at50%, 75% and 90% of budget
    • At the limitYour apps stop

  • Up to 32 vCPU and 32 GB

    Per service on Pro, 8 vCPU and 8 GB on Hobby. The change goes out with the next deploy.

  • Deploys without downtime

    The running version keeps serving until the new one passes its healthcheck. A service with a volume is the exception: it restarts with a short gap.

  • Restart policies

    Restart always, only on failure, or never. A crashed replica comes back on its own.

Start small. Grow when it’s busy.

Start free Read the quickstart
Paste into your terminal, with the token from your dashboard
$ claude mcp add --transport http dawnrose \
    https://app.dawnrose.ai/mcp \
    --header "Authorization: Bearer YOUR_TOKEN"