How Polaris Mesh Redefines Service Discovery and Governance for Cloud‑Native Apps

Background

Polaris Mesh is an open‑source service‑discovery and governance platform released by Tencent Cloud on 8 September. Internally it has registered over 1 million services, more than 5 million instances and handles over 30 trillion API calls per day.

Motivation: From Monolith to Microservices

Full‑system recompilation and redeployment for any change → high risk and slow iteration.

Defects in a single module can affect the whole application, reducing availability.

Uneven request load prevents targeted horizontal scaling of hot modules.

Service‑Discovery and Governance Approaches

Spring Cloud family integrates many components but lacks a unified data and control plane and is hard to manage across multiple languages.

Kubernetes Service registers instances in etcd and uses DNS for discovery, but provides no governance features.

Istio‑style service mesh hijacks traffic to offer discovery and governance with low code intrusion, at the cost of additional resource consumption and operational complexity.

Polaris Mesh Overview

Polaris Mesh builds on a high‑availability registration core and adds governance capabilities. It supports multiple protocols, health‑checking, and can be deployed in containerized (Kubernetes) or sidecar‑based mesh environments.

Key Features

Registration & Discovery : High‑capacity, multi‑protocol registration with health checks.

Traffic Scheduling : Dynamic routing (region, canary, isolation) and load‑balancing algorithms (weight‑random, least‑load, consistent‑hash).

Circuit Breaking & Degradation : Instance‑, interface‑ and service‑level policies.

Access Control : Authentication rules and rate‑limiting (per‑caller or global).

Service Mesh : Unified control and data planes; SDK mode (language‑specific, non‑intrusive) and Sidecar mode (traffic hijacking).

System Architecture

The platform consists of:

Core : Console, control plane (registration + governance) and data plane.

Ecosystem : Integration modules for frameworks, gateways and Kubernetes.

Data‑plane implementations:

SDK mode : Language‑specific libraries that can be embedded in application code without modifying request flow.

Sidecar mode : A sidecar proxy that intercepts traffic, suitable for container‑native deployments.

Best Practices

Use Polaris as a unified service‑discovery layer across all services, supporting million‑scale client access and multi‑center deployment for disaster recovery.

Choose SDK mode for framework‑centric services to avoid additional operational overhead; use Sidecar mode for container‑native workloads where traffic hijacking is acceptable.

Integrate Polaris with API gateways to forward requests directly, enabling gateway‑level service‑mesh capabilities.

Combine traditional registration‑center solutions with Kubernetes‑based meshes to provide a single governance layer for both VM and container environments.

Open‑Source Roadmap

Enhance discovery and governance features.

Expand language SDKs for the data plane.

Optimize Sidecar performance.

Strengthen integration with related open‑source components.

Improve documentation and community contribution processes.

Source code and issue tracking are available at https://github.com/polarismesh/polaris.