Why Gritiva
Why Gritiva#
A short answer to "why does this exist when I already have Kubernetes / Tailscale / Coolify / Docker Swarm?"
The thesis in one sentence#
You should be able to run apps on a fleet of cheap VMs the same way you run them on a single laptop — no cluster, no Kubernetes, no learning a new mental model — but with mesh networking, isolation, and a panel so you don't ssh into every box.
What we are#
Gritiva is infrastructure software for the people Kubernetes leaves behind: indie hackers, small ops teams, internal IT, agencies running 5-50 customer environments. Three primitives:
- Scope — a Linux network namespace + cgroup. Lighter than a VM, more isolated than a container, no cluster required.
- gmesh — WireGuard between scopes across VMs. No central relay. No service mesh sidecar.
- Panel — one web UI to provision a VM, attach domains, share services, and watch metrics across the whole fleet.
That's it. No etcd. No control-loop fighting. No manifests.
When NOT to use Gritiva#
Be honest first.
- You're already on Kubernetes and it's working — don't migrate. We don't replace k8s; we replace the pre-k8s "I have 6 VMs and no idea what's where" problem.
- You need >100 nodes per cluster — gmesh's full-mesh topology hits practical limits around 50 peers per VM. We're working on it (BGP-style route reflectors), but today, we are not your tool.
- You need every pod to autoscale on CPU — we don't have an autoscaler. Scopes are static-ish; you start them, they run.
If those are dealbreakers, use Kubernetes (or Nomad, or just plain ECS).
When Gritiva IS the right tool#
- You have between 2 and 50 VMs, mostly long-running web apps, databases, and background workers.
- You want per-app network isolation without learning Cilium, Calico, or a service mesh.
- You want one place to attach a domain and have HTTPS just work, without setting up Caddy / Traefik / nginx-ingress yourself on every box.
- You want mesh between VMs in different clouds without buying SD-WAN.
- You want to read the source of the thing managing your infrastructure.
That's the wedge. If that's you, try it free →
How we compare#
We're not going to pretend we beat every competitor at their own game. Here's the honest table.
| Gritiva | Tailscale | Coolify / Dokploy | Kubernetes | |
|---|---|---|---|---|
| Per-VM mesh | ✅ | ✅ | ❌ | (CNI does it) |
| Per-app netns/cgroup isolation | ✅ | ❌ | ❌ | (per-pod) |
| One panel to manage many VMs | ✅ | (partial) | ✅ | (need rancher/lens) |
| Domain → app routing built in | ✅ | ❌ | ✅ | (ingress controller) |
| Source-available | ✅ | (partial) | ✅ | ✅ |
| Cluster-required | ❌ | ❌ | ❌ | ✅ |
| Steady-state CPU/RAM overhead per node | <100 MB | ~20 MB | ~200 MB | ~1 GB |
| Learning curve | days | hours | days | months |
We're closest to Tailscale + Coolify smashed together, with a thinner application abstraction (scope) than Coolify's docker-compose unit, and stronger isolation than Tailscale alone gives you.
The team's bet#
We think the next decade of infra is "how do I run a small fleet of long-lived workloads with the operational ergonomics of a managed PaaS but the price of a Hetzner dedicated server". Gritiva is our answer to that question.