David is a Director of Software Engineering at Pacific Biosciences, a maker of DNA sequencers, and he has been working as a manager and engineer for 28 years. He has a long time interest in faster and more regular software releases for hardware, and recently when and when not to apply DevOps principles to embedded software. Previously a manager and engineer at Apple, he was responsible for Mac software updates at a time when quarterly releases was a big improvement. At PacBio, he moved internal releases to three times per day.
I've been told DevOps doesn't apply to IoT or to embedded systems. Customers don't want their devices continuously updating. While that may often be true, we're seeing companies like Tesla turn frequent software updates with smaller batch sizes of features into a competitive advantage. At PacBio, I was curious if these same principles could be applied to our DNA sequencers. Over the course of nine years and three hardware platforms, we increased our cadence by a factor of 20x and significantly reduced our time to resolve issues. We learned a ton along the way with some unexpected wins, which I'd like to share in this talk.