Linkerd 2.15: Service Mesh Observability and Performance

Introduction

Linkerd’s appeal has always been pragmatic: give me mTLS, metrics, and sane defaults—without turning every node into a sidecar tax collector. The interesting part of “lightweight” meshes is what they choose not to do, and how they still deliver enough policy and visibility for production.

Linkerd 2.15, released on July 20, 2025, tightens that balance with improved observability, faster data plane behavior, stronger policy ergonomics, and more Gateway API support—while keeping the operational model intentionally simple.

Why this matters in practice

• Better signals, less guessing: richer metrics/tracing help answer “is it the mesh or the app?” faster.
• Policy you’ll actually use: improved enforcement + validation reduces the “policy exists but nobody trusts it” pattern.
• Performance headroom: efficiency gains matter most at scale, where “small overhead” becomes a monthly bill.
• Gateway API support: helps teams modernize routing without replatforming everything at once.

Enhanced Observability

• Metrics expansion provides more detailed service-level metrics for traffic analysis.
• Distributed tracing improvements enable better correlation of requests across services.
• Service profiles enhancements provide better visibility into service behavior and performance.
• Dashboard improvements offer more intuitive visualization of service mesh metrics.

Performance Improvements

1. Latency reductions minimize the overhead of service mesh operations.
2. Throughput enhancements improve handling of high-traffic workloads.
3. Resource usage optimizations reduce CPU and memory consumption.
4. Connection pooling improvements enhance efficiency for service-to-service communication.

Policy Enforcement

• Server authorization provides fine-grained control over service-to-service communication.
• Network policy integration enables consistent policy enforcement across networking layers.
• mTLS improvements enhance security with better certificate management.
• Policy validation provides better feedback on policy configuration errors.

Gateway API Support

• HTTPRoute support enables Gateway API-based routing configuration.
• Traffic splitting provides sophisticated traffic management capabilities.
• Backend selection improvements enable better service selection and health checking.

Upgrade and Identity Notes

• Control plane first: linkerd upgrade | kubectl apply -f - then linkerd check --proxy before touching workloads.
• mTLS identity rotation: Rotate trust anchors/issuer with linkerd identity rotate-identity-issuer and ensure pod restarts are staggered.
• Proxy CPU/memory: Plan ~50–80m CPU / 80–120Mi per proxy at baseline; set proxy resources via linkerd inject --proxy-cpu/--proxy-memory.
• Gateway API: Install the linkerd-gateway extension if you use HTTPRoute; keep IngressController only where needed.
• Dual stack: Verify enableDualStack is set before upgrade if the cluster already uses IPv6 addresses.

Getting Started

# Install Linkerd CLI
curl --proto '=https' --tlsv1.2 -sSfL https://run.linkerd.io/install | sh

# Install Linkerd
linkerd install | kubectl apply -f -
linkerd check

Summary

Linkerd 2.15 continues to provide the simplest and most efficient service mesh for Kubernetes, making service-to-service communication secure, observable, and reliable.