Virtualization Technology News and Information
Article
RSS
Service Mesh Benchmarking Results: Linkerd Dramatically Outperforms in Cost and Performance

The maintainers of Linkerd announced the results of the latest benchmark analysis of service mesh performance. These tests, designed to measure mesh performance under real-world conditions, showed that Linkerd consumed an order of magnitude less memory and CPU and introduced up to 400% less latency than Istio at scale.

"My team and I first deployed Istio because it seemed like the popular option, but we found it hard to use, resource-heavy, and too complex," said Chris Campbell, cloud platform architect, HP. "In contrast, Linkerd worked right out of the box, consumed a fraction of the system resources and had an understandable architecture. These test results are not surprising -Linkerd makes it obvious to adopt service mesh and delivers the best performance."

The benchmarks evaluated the latest versions of Linkerd and Istio and reported the relative costs of each mesh against the baseline condition of no service mesh. These costs included both the CPU and memory consumed by the mesh as well as the user-facing latency the meshes introduced.

At the highest traffic level measured, the benchmarks showed:

  • Latency:Istio introduced 36% more additional median latency and 438% more additional maximum latency than Linkerd.
  • Control plane:Istio's control plane required 2.5x the memory and over 50x the CPU that Linkerd's control plane required.
  • Data plane:Istio's Envoy proxies consumed 8x the memory and 8x the CPU that Linkerd's Rust-based proxies did.

"The data plane is the most critical part of the service mesh. This is the part of the mesh that scales out, and also the part that imposes a user-facing latency cost," said William Morgan, CEO of Buoyant and one of the creators of Linkerd. "The tremendous difference in performance and resource cost between Linkerd and Istio largely comes down to Linkerd's Rust-based 'micro-proxy' implementation on the data plane. These micro-proxies not only make Linkerd significantly easier to operate than Envoy-based service meshes like Istio, they enable it to deliver tremendous gains in efficiency and performance. These numbers speak for themselves."

The benchmark tests followed a 2019 analysis performed by Kinvolk which showed a similar performance gap between the meshes. Today's tests use the same Kinvolk benchmark suite with latest stable releases of both projects (Linkerd 2.10.2 and Istio 1.10.0) and run on a CNCF Kubernetes cluster provided by Equinix Metal. While the 2019 Kinvolk tests evaluated performance at 500 requests per second (RPS) and 600 RPS, these tests evaluate the service meshes at over three times the scale at 2,000 RPS. (The full discussion of experimental methodology and results are available in the project's blog post about the benchmarks.)

Linkerd's focus on performance, security, and simplicity has led to a 300% increase in downloads over the past year. Linkerd's adopters range from innovative startups to large international enterprises and include Microsoft, H-E-B, EverQuote, HP, the Nordic electronics retailer Elkjøp, and Entain, the global sports betting and gaming operator, which recently reported reducing server load by over 50% while increasing request volume by 10x with Linkerd. Linkerd was recently recognized as the Best Open Source DevOps Tool of 2020 by the Tech Ascension Awards for its ability to address core challenges faced by engineers building and operating modern applications.

What end users have to say about Linkerd

"Like many organizations, we considered Istio. But our research led to the conclusion that we would need a team of developers just to run it. It was too complicated, requiring ongoing, active attention-it's not fire and forget. We looked at other solutions and ended up with a shortlist of half a dozen different options, but the one that stood out was Linkerd." -- Steve Reardon, DevOps Engineer, Entain

"We've felt encumbered by [Istio's] complexity every time when configuring, maintaining or troubleshooting in our clusters. Our suspicions were that since we hardly used any of the capabilities, we could probably make do with a much simpler alternative. So, after yet another 'Oh... This problem was caused by Istio!'-moment, we decided the time was ripe to consider the alternatives." - Frode Sundby, Senior Engineer at NAV

Published Thursday, May 27, 2021 9:29 AM by David Marshall
Filed under:
Comments
There are no comments for this post.
To post a comment, you must be a registered user. Registration is free and easy! Sign up now!
Calendar
<May 2021>
SuMoTuWeThFrSa
2526272829301
2345678
9101112131415
16171819202122
23242526272829
303112345