Skip to content
Docs
Why np

Why np

The Gap

Most deployment tools cluster at two extremes:

Tool Scale Band Problem
Docker Compose 1-3 servers No health checks, no multi-node scheduling, manual restarts
Kubernetes 30+ servers Operational complexity, resource overhead, steep learning curve
Bare Nomad 5-500 servers You assemble Consul, Traefik, monitoring, logging yourself
np 3-100 servers Purpose-built for this band

The gap is real: teams with 3-15 servers running 20-200 services have no tool that gives them production-grade capabilities without production-grade complexity.

Design Principles

1. Battery-included, not reinvented

np doesn’t write a scheduler. It doesn’t write a service mesh. It integrates the best open-source components — Nomad for scheduling, Consul for service discovery, Traefik for ingress, OpenObserve for observability — into one coherent platform.

# No assembly required
np init       # Sets up Nomad + Consul + Traefik + OpenObserve
np deploy app # Ship it

2. One file is enough

No CRDs. No Helm charts. No Terraform providers. One np.yaml describes your entire deployment.

name: api
binary: ./api-server
health_check: /health
resources:
  cpu: 200m
  memory: 128mb

3. CLI-first, always

The CLI is the product. No web dashboard you have to learn. No API you have to script against.

np deploy → np status → np logs → np scale

Every operation fits in 30 seconds.

4. Runs on anything

Raspberry Pi. Old server. Your laptop. Cloud VM. np treats them all the same.

  • Single binary (< 20 MB)
  • No kernel modules
  • No container runtime required (binaries run natively)
  • 100 nodes, 500+ services without tuning

5. Honest about boundaries

np doesn’t pretend to replace Kubernetes at scale. It doesn’t try to be everything.

What np is NOT

  • ❌ Not a Kubernetes alternative. np targets 10-100 nodes. K8s targets 100-10,000. Different category.
  • ❌ Not a Nomad competitor. np uses Nomad as its scheduler. It adds the missing pieces.
  • ❌ Not infrastructure-as-code. No Terraform providers, no CRDs. Configuration, not code.
  • ❌ Not a PaaS. np runs on your machines. No vendor lock-in. No per-deployment pricing.
  • ❌ Not for every team. If you have a dedicated platform team running K8s and love it, stick with it. np is for the other 95%.

Next: Comparison with alternatives →

Getting StartedComparison