The Cloud Native Computing Foundation has launched version 2.0 of Linkerd, which provides service discovery, routing, failure handling, and visibility for applications. This release introduces a more incremental scale and features “improvements to performance, resource consumption, and ease of use,” the company said in the announcement.
The company explained that the project has been reconfigured as a “service sidecar” rather than a cluster-wide service mesh, a change that gives more finely-detailed configuration options to developers and provides “automatic observability, reliability, and runtime diagnostics without configuration or code changes” on individual services rather than requiring installation across the whole cluster.
“With the 2.0 release, the community focused heavily on the idea of ‘service ops,’ whereby service owners are responsible for not just for building their service, but also deploying it, maintaining it, and waking up at 3 am if it breaks,” Oliver Gould, core maintainer of Linkerd and CTO of Buoyant, said in the announcement. “Service owners are the ultimate customers of all this platform technology we’re building, and we wanted to address their needs directly.”
The open-source project is maintained by a community including developers from companies such as Salesforce, Walmart, Comcast, CreditKarm, PayPal, WePay and Buoyant and can be found at the project’s GitHub repository. The company highlighted the following features in its announcement:
- A self-contained “service sidecar” design that augments a single service without requiring cluster-wide installation.
- An incremental path to cluster-wide service mesh, whereby service sidecars across multiple services link to become a service mesh.
- A zero-config, zero-code-change installation process.
- Automatic Grafana dashboards and Prometheus monitoring of service “golden metrics.”
- Automatic TLS between services, including certificate generation and distribution.
- A complete proxy rewrite in Rust, yielding orders of magnitude improvement in latency, throughput, and resource consumption.
“Before Linkerd 2.0, for my services, all I had was statistics for my public API. Now I can see on a very granular level how each of my services are behaving,” Pascal Bourque, CTO and co-founder of Studyo, a task and project manager designed for schools, said in the announcement. “For me this is gold. The fact that it is painless to install is even better.”