HAProxy

## What it is

HAProxy is a TCP/HTTP load balancer and reverse proxy written in C. In Kubernetes contexts it shows up most often as an Ingress controller (via the community `haproxy-ingress` or Kubernetes-native `haproxy-ingress-controller` projects) or as an external LB fronting a cluster's API server or workloads. Core description isn't published in the data available here, but the codebase and category placement (Orchestration & Management) confirm it's used as traffic-routing infrastructure inside or in front of clusters, not as a scheduler or control-plane component itself.

## Who builds it and why

428 contributors on a single C repo is a substantial number for infrastructure software — this isn't a side project. HAProxy predates Kubernetes by well over a decade and has an established maintainer base and commercial entity (HAProxy Technologies) behind it, though founding date and CNCF status aren't confirmed in the data provided. The project keeps shipping: a commit dated as recently as mid-2026 in this dataset signals active maintenance, not a project coasting on past work. The motivation is straightforward — HAProxy exists because load balancing correctness and performance at scale is a hard, narrow problem, and it's been solving that problem since before "cloud native" was a phrase.

## Production readiness signal

6,661 GitHub stars is modest for a project this old and this widely deployed — HAProxy's actual footprint (embedded in cloud provider LBs, CDNs, and enterprise networking stacks) massively exceeds its GitHub visibility, because most users interact with it as infrastructure, not as a repo they star. 428 contributors and continuous commits are strong signals of institutional investment and real-world battle-testing. The license field shows `NOASSERTION`, which means GitHub's license detector couldn't confidently classify it — worth manually verifying the actual license terms before you build compliance assumptions around it; don't take "open source" for granted here without checking. No published release version or founding date in this dataset is a gap you should fill via haproxy.org before making procurement decisions.

## Who should use this

Teams that need battle-tested L4/L7 load balancing with fine-grained control over routing, health checks, SSL termination, and connection handling — especially where NGINX's config model or Envoy's operational complexity don't fit. Good fit if you already run HAProxy outside Kubernetes and want config/tooling consistency inside the cluster, or if you need HAProxy-specific features (stick tables, advanced ACLs, low-latency L4 proxying) that generic Ingress controllers don't expose cleanly.

## Who should NOT use this

If you're starting fresh on Kubernetes with no existing HAProxy investment, the default Ingress-NGINX or a service mesh (Istio, Linkerd) with built-in ingress will have better community support in a Kubernetes-native context and more Kubernetes-specific documentation. If you need service-mesh features (mTLS, fine-grained traffic shifting, observability baked in) rather than pure load balancing, HAProxy alone won't get you there — you'd be bolting on additional tooling. Teams uncomfortable with C-based software's operational model (config reloads, binary upgrades) versus more "cloud-native" Go-based alternatives should also weigh that friction. And given the `NOASSERTION` license flag, legal/compliance-sensitive organizations should resolve licensing clarity before adoption, not after.

## Alternatives

- **NGINX / NGINX Ingress Controller** — more common default choice in Kubernetes, larger ecosystem of Kubernetes-specific tooling and docs.
- **Envoy** — built for cloud-native environments from day one, the default data plane for most service meshes (Istio, Consul Connect).
- **Traefik** — Kubernetes-native by design, automatic service discovery and simpler config for dynamic environments.

## Pricing

HAProxy the core proxy is open source. HAProxy Technologies also sells HAProxy Enterprise with support, additional modules, and a management UI — if you're evaluating for production, check current terms directly with them since specifics aren't in the data available here.