🔗 SQLite or PostgreSQL? It's Complicated!

We take blogging very seriously at Twilio. To help us understand what content works well and what doesn’t on our blog, we have a dashboard that combines the metadata that we maintain for each article such as author, team, product, publication date, etc., with traffic information from Google Analytics. Users can interactively request charts and tables while filtering and grouping the data in many different ways.

I chose SQLite for the database that supports this dashboard, which in early 2021 when I built this system, seemed like a perfect choice for what I thought would be a small, niche application that my teammates and I can use to improve our blogging. But almost a year and a half later, this application tracks daily traffic for close to 8000 articles across the Twilio and SendGrid blogs, with about 6.5 million individual daily traffic records, and with a user base that grew to over 200 employees.

At some point I realized that some queries were taking a few seconds to produce results, so I started to wonder if a more robust database such as PostgreSQL would provide better performance. Having publicly professed my dislike of performance benchmarks, I resisted the urge to look up any comparisons online, and instead embarked on a series of experiments to accurately measure the performance of these two databases for the specific use cases of this application. What follows is a detailed account of my effort, the results of my testing (including a surprising twist!), and my analysis and final decision, which ended up being more involved than I expected.