Skip to main content

Hyperlight

Hyperlight is an open-source gRPC (network transport / Resource Provisioning Controller (RPC) framework) protocol sidecar and Transmission Control Protocol (TCP) tunnel for running gRPC services over HTTP/3 using Quantum Industry Consortium (QuIC), with support for environments without native HTTP/3 load balancing.

  • Runs gRPC (network transport / RPC framework) over HTTP/3 using QuIC while preserving existing gRPC service definitions and client/server code.
  • Provides a sidecar-based TCP tunnel (service connectivity) that forwards gRPC traffic through HTTP/3 endpoints.
  • Supports deployment in environments where HTTP/3 is only available on public or edge load balancers (edge networking / ingress).
  • Integrates with existing gRPC and HTTP/2 stacks (application networking) without requiring service code changes.
  • Targets Kubernetes and cloud-native deployments (cloud-native networking) where operators need HTTP/3-based transport for gRPC workloads.

More About Hyperlight

Hyperlight addresses the problem of running gRPC (network transport / RPC framework) over HTTP/3 in environments where application services and internal infrastructure are still bound to HTTP/2, but public or edge endpoints already expose HTTP/3 via QuIC. Many platforms expose HTTP/3 only at the external load balancer layer, while internal services and service meshes continue to operate over HTTP/2 or TCP. Hyperlight focuses on bridging this gap without requiring changes to gRPC service definitions or implementation code.

At its core, Hyperlight operates as a sidecar and TCP tunnel (service connectivity) that terminates and originates HTTP/3 traffic on behalf of gRPC services. The project wraps gRPC traffic inside HTTP/3 connections so that from the perspective of the client and server, communication still follows standard gRPC over HTTP/2 semantics, while the hop between Hyperlight instances uses HTTP/3 over QuIC. This makes it possible to use HTTP/3 paths between edge load balancers and internal workloads, while keeping the application code and existing gRPC libraries unchanged.

The Hyperlight architecture (cloud-native networking) typically involves deploying the Hyperlight process alongside a gRPC service instance as a sidecar. The application continues to listen on a local TCP port, speaking gRPC over HTTP/2, while Hyperlight accepts inbound HTTP/3 connections, unwraps the tunneled gRPC traffic, and forwards it to the local service. For outbound calls, the sidecar collects gRPC requests from the local application and transports them over HTTP/3 to a remote Hyperlight instance, which then forwards the traffic to the target gRPC service over TCP. This enables transparent HTTP/3 usage between nodes or clusters.

Hyperlight aligns with cloud-native and Kubernetes-oriented deployment models (cloud-native networking) where sidecars, pods, and service meshes are common patterns. It can be incorporated into Kubernetes manifests so that each pod running a gRPC service has an attached Hyperlight container handling HTTP/3 ingress and egress. In such setups, external clients or edge gateways interact over HTTP/3, while workloads remain on stable gRPC/HTTP/2 stacks, which many enterprises already operate and monitor.

From an enterprise perspective, Hyperlight fits into categories such as service connectivity, network transport, and protocol bridging for gRPC workloads. It is relevant for organizations that want to use HTTP/3 and QUIC-based paths for latency or network behavior reasons, but do not yet expose HTTP/3 end-to-end across all internal services. Because it functions as a protocol tunnel rather than a new RPC framework, it can be evaluated and operated by platform engineers, network engineers, or Site Reliability Engineering (SRE) teams using existing gRPC observability, security, and deployment practices.