Skip to main content

API Benchmarks - Why it's Important, and How

Matteo Figus (OpenTable)
Operations
Beekman
Average rating: ***..
(3.80, 5 ratings)
Slides:   1-PDF 

As developers we rely on technologies that, with a minimum effort, can guarantee some pretty decent results in terms of performance. Modern web frameworks handle concurrency and thread management without requiring much plumbing code. Sophisticated and relatively cheap cloud services help us monitor our applications, providing dashboards, reports, and alerting systems. Deploying on the cloud we can run our services and even auto-scale them depending on how much power we need. Even with these tools we must still own the responsibility of writing good quality code, testing it properly, and deploying as fast as possible in order to optimize the delivery process of our products.

But what about performance? I mean, what about the relationship between the code we write every day, and the way we impact overall performance? Are we sure that we are not deploying to production something that is degrading our services’ performance?

During the last year I’ve been experimenting with performance monitoring, specifically on REST APIs. This talk is about my story: why I started benchmarking one of my APIs, how I automated that process in order to make it part of my continuous integration environment in each of my projects, and some lessons I learned.

Photo of Matteo Figus

Matteo Figus

OpenTable

I’ve been writing software for the last 15 years, at least half as a professional. Now I am based in London and I work at OpenTable, where I like to share my time between writing software, testing it, deploying it properly, and monitoring it. Main interests are node.js, C#, the open-source world, and the dev-ops movement.