“Every market eventually becomes a two-horse race” — so say Ries and Trout in their 22 Immutable Laws of Marketing. It’s Coke and Pepsi, Crest and Colgate, McDonald’s and Burger King. In the world of container orchestration, it’s Swarm and Kubernetes, each vying for the top spot, each working to prove its superiority in performance and simplicity.
Now “the world of container orchestration” has early-stage metrics on both, thanks to a third-party study by container blogger, Jeff Nickoloff. And, to put it plainly, Swarm knocks Kubernetes out of the park.
We’re talking about:
- 5x faster container startup and 7x faster delivery of operational insights, with Swarm often starting containers in under .5 seconds, (while Kubernetes often required 2 seconds or more), and Kubernetes taking up to 7x longer to list running containers as cluster approached full load.
- A Swarm system architecture that’s greatly simplified (relative to Kubernetes), with 8x fewer “hops” needed to execute simple commands like “run” or “list.”
It’s important because simplified architecture means fewer potential points of failure, fewer things to learn, and decreased support costs.
And though Kubernetes’s inferior performance may be “acceptable under today’s early phase of container implementation,” the continued proliferation of containers will make Swarm’s superior performance “more and more relevant” over time (as InformationWeek’s Charles Babcock has written).
At this early stage, as customers choose the first tools they’ll learn and invest in, Nickoloff’s performance benchmark has labeled Docker Swarm the clear winner. Between the 5x performance advantage, the ability to manage environments under load, and the core simplicity (which makes Swarm easier to troubleshoot and learn), Swarm impresses as the best tool currently on hand. For learning orchestration today, and for orchestrating at scale tomorrow, we know of no better alternative.
Check out the link below for Docker’s writeup on Jeff Nickoloff’s Swarm vs. Kubernetes benchmarking: