The latest edition of The New Stack @ Scale Podcast, sponsored by New Relic, takes a dramatically new approach. The show kicks off with a special segment from The New Stack contributor Scott Fulton III that incorporates interviews with Avi Cavale of Shippable, Apcera founder and CEO Derek Collison, and Martin Croker, DevOps Capability Lead at Accenture, into a deep look at the interaction of platforms, containers, and continuous integration / continuous deployment (CICD). One key, Scott discovers, is to make sure everyone in the organization is on board, which means making sure IT, developers, and business leaders “see eye to eye to eye” (time code 1:41).
It’s a fascinating investigation, and it’s a great springboard for the meat of the podcast, where host Alex Williams and I are joined by Google Staff Developer Advocate Kelsey Hightower.
The conversation covers a wide variety of important topics, but I was particularly interested in these collaboration issues, and the idea that you need to define and document APIs even for internal applications in order to scale. As Kelsey put it, “When you build … an application meant for internal delivery, there’s often not a lot of documentation, there aren’t many APIs that are meant to be consumed by other people in the organization, like a third party would. You end up with a bunch of intimate knowledge contained within very small groups of people.”
It makes more sense, Kelsey added, “to treat those applications as if they were going to be consumed by a third party…. They need documentation, they need APIs, they should be able to be consumed without talking to the team that built them” (14:29).
There’s nothing wrong with intergroup collaboration, Kelsey added, but he wondered whether embedding people across teams is truly scalable. You can reduce issues and mistakes due to human interactions with cross-functional teams including documentation people, QA people, and systems people working to make sure that the APIs can be properly utilized.
But that’s only the beginning. Kelsey also touched on the evolution of SLAs into API management (25:08) and the connection between scale and complexity and the ability to abstract that complexity with the right platform that fits your needs (27:56). I also appreciated Kelsey’s insight into the meaning of performance: “The biggest problem I see is when someone says, ‘I think it’s slow.’ When someone starts the conversation with, ‘I think it’s slow, I think the database is slow, I think the email service is too slow’… that just goes in weird directions.” DevOps, he added, has the potential to help organizations decide what SLA is reasonable (22:59).
Finally, Kelsey addressed the role of containers in thinking about platform management: “The simple act of putting something in a container means very little without talking about the target platform they’re going to run on” (30:48).
All this and much more in the latest episode of The New Stack @ Scale Podcast: APIs Alone Do Not Tear Down the Silo.