Solving Orchestration with Joyent’s Casey Bisson

Hollywood often misrepresents technology. has covered this subject before, and Joyent’s Casey Bisson gave us still more examples during his Container Summit talk earlier this month.

Casey discussed infrastructure-independent applications that scale with ease (Terminator 3), and a “cross-platform” virus that works on alien technology (Independence Day) – to the amusement of his Summit audience. But Bisson doesn’t think his examples are impossible. To the contrary: he believes we can learn from Sci-Fi films that depict code as “self-assembling, self-aware, self-healing, infrastructure independent, and scheduler independent” – code that needs “no external orchestrators.”

It’s important because “orchestration breaks a lot of our applications,” according to Bisson, who likens the role of a Site Reliability Engineer (SRE) to “Bruce Lee playing ping pong with nunchucks.”

“Orchestration is tough,” Bisson said. “SREs really are the Bruce Lees of the modern era and they’re doing this beautiful thing playing ping pong with all of the wrong tools… Perhaps that makes it easier to understand why we suffer so many problems.”

According to Bisson, the breakdown happens due to a separation between dev and ops. Developers are creating apps independent of the orchestration layer – apps that “work on their machine” – and leaving it to an SRE to handle “the orchestration and the integration for production.” Devs aren’t “running the orchestration on their laptops when they’re doing development” and often “don’t even have access to see what’s going on in the orchestration.”

Bisson says the essential question should not be “does it work on my machine,” but “does it orchestrate on my machine?” And the follow-up: “does it orchestrate the same way in production?”

To that end, he and his team built an application that handled orchestration without relying on any external orchestrators. They were able to do this with both a simple modern application and a more complex app with a classic MySQL database.

“We realized there’s an approach that we’re taking that actually allows us to make devops site reliability a repeatable pattern and bring it to the developers,” Bisson said. “None of this is using a big external orchestrator. It works on my laptop the same way as it does in the cloud.”

Please follow the link below to view Casey Bisson’s full presentation. His examples are open source so you can try them yourself:



About the Author: