XLB is a packet-level load balancer that forwards traffic like a router, not a proxy. Built for workloads handling millions of requests per second—deploy on bare metal or Kubernetes as a portable alternative to AWS NLB or GCP Network Load Balancer.
Traditional L4 load balancers like HAProxy and NGINX are reverse proxies—they terminate connections and create new ones to backends. This adds latency, limits throughput, and requires constant tuning to perform at scale.
XLB operates at the packet level, forwarding traffic directly like a router or hardware load balancer. It's L4-aware (understands TCP/UDP flows) but doesn't terminate connections. The result: near-zero latency overhead and the ability to saturate network links.
Best of all, XLB is deploy-and-forget—no endless config tweaking or performance tuning. It just works.
XLB combines the performance of hardware load balancers with the flexibility of software.
Kernel-bypass packet processing achieves line-rate forwarding with minimal CPU overhead.
Graceful connection draining and hot config reloads keep services available during changes.
First-class LoadBalancer and Gateway API support. Works with any CNI and autoscaler.
Export metrics and traces to any OpenTelemetry-compatible backend like Datadog, Grafana, or Prometheus.
Single binary deployment with systemd. No dependencies, no container runtime required.
No endless config tweaking or performance tuning. XLB just works—without the maintenance overhead of traditional proxies.
How XLB compares to cloud and software load balancers
XLB runs wherever your workloads run
Tell us about your use case and we'll set up an evaluation for your infrastructure.