All of the talks I’ve given live the lisa/conferences repository. Within that repository I have links to slides and any recordings. That repository will be the authoratative source for links and details; on this blog I’ll expand or give some background, where available.
In late 2015 I was invited by a coworker (@ahalliop) to give a talk at a local meetup she helps organize. At the time, we were creating a new server environment for PCI compliance. The previous, non-PCI environment, had grown very organically over time and had proven to have a number of pain points.
This talk originated from wanting to share some lessons from my team at FreshBooks coming up with the first SLOs. It wasn’t an easy process because they were put into an existing monitoring ecosystem. We were trying to get buy-in for service ownership (the idea that people who wrote the software would be responsible for it, instead of “ops team” de facto being responsible). While that push never panned out while I was working there, the ops team did pilot a project to come up with some SLOs for MySQL and RabbitMQ.
The talk focuses on RabbitMQ because there are some more lessons there. One of those lessons is that the customer may think your service is the best source of truth for an objective, and that might be incorrect. Talking with the customers to understand the goals and best source of data is important, as is iterating on the SLO-creation process.
To date, this talk has been given twice.
I had mentioned to Jennifer Davis, co-author of Effective DevOps: Building a Culture of Collaboration, Affinity, and Tooling at Scale, my talk topic and she expressed interest. To fill a speaker slot at the last minute, she graciously invited me to present at the two-day conference.