Greatest Hits Archives - Software Engineering Daily · Greatest Hits Archives - Software Engineering Daily

Infrastructure Wars with Sheng Liang

·54 min·3 clips
Rancher 2.0 required throwing away almost all of the 1.x code.
1. Software Engineering Daily's episode "Infrastructure Wars with Shang Liang" follows Rancher Labs' view of Kubernetes, cloud infrastructure, and platform battles. 2. The Software Engineering Daily host speaks with Shang Liang, the former lead developer on the original Java Virtual Machine and CEO of Rancher Labs, whose career links Java, CloudStack, OpenStack, and Kubernetes. 3. The episode asks whether technology wars are necessary and whether companies can survive when competing standards force them to rewrite products. 4. Liang says his first job out of school was building the JVM, where the team had to move beyond an interpreter by adding just-in-time compilers and code generators. 5. He recalls that an early Java runtime could be crashed by a simple program within minutes, which is why the work still felt unfinished when he joined after graduating in 1996. 6. Liang says Java's timing lined up with the internet boom and the Y2K scare, when companies were rewriting applications and looking for a new platform. 7. He argues that Java became more successful than the industry's early expectations because open source, standard implementations, and compatibility eventually mattered more than proprietary variants. 8. On OpenStack and CloudStack, Liang says the founding Rancher team learned that cloud infrastructure was not a commodity because AWS, Google, Azure, and vSphere all exposed different APIs and services. 9. He says that lesson led Rancher toward Docker and Kubernetes, where packaging and deployment could finally work across Google, Amazon, Azure, vSphere, and bare metal. 10. Liang describes the container-orchestration wars as the period when customers paused buying decisions because they did not know which technology would win. 11. He says Apache Mesos and Marathon worked well in practice but were still incomplete, which forced Rancher to do a lot of extra work on top. 12. Rancher built its own Docker orchestration layer called Cattle, and Liang says many people still remember it because the market had so few stable options. 13. He says the later Kubernetes migration was so disruptive that Rancher 2.0 had to throw away almost all of the 1.x code and start over around Kubernetes. 14. Liang argues that competition can improve products and raise awareness, but it also creates casualties when companies bet too hard on the wrong standard. 15. He compares current service-mesh competition to earlier wars, saying Linkerd has the openness of the CNCF while Istio has more features and Google's credibility. 16. He says Rancher supports Istio today through its Reel platform, but the SMI effort could make Linkerd or another implementation a plugin later. 17. The conversation becomes more pragmatic when Liang says Kubernetes should be treated like Linux: hidden underneath the product, but central to edge deployments from Chick-fil-A stores to wind turbines and oil platforms. 18. The interview stays measured and reflective, with Liang returning to Rancher, Kubernetes, compatibility, and product-market fit whenever the market history gets abstract. 19. Best for platform engineers, Kubernetes operators, and Rancher Labs followers. 20. Skip if you want a light, nontechnical interview about Kubernetes.

As heard by us

A candid look at how competing container standards can force a software company to rebuild its foundation.

Infrastructure wars turn into a company-survival story as Rancher Labs' CEO traces how competing container orchestrators pushed the business toward a near-fatal reset.

Read the full review in PlayNext →

Why you'd press play

Press play when you want the cautionary tale of Rancher surviving the container orchestration wars and rebuilding around Kubernetes.

Read the full recommendation in PlayNext →
Listen to the show on