written rows per day
processed per month
requests per day
Vercel takes care of the hard things for developers who want to build fast websites—helping them deploy instantly, scale automatically, and serve personalized content. With an obsessive focus on end-user performance, Vercel is the best place to deploy any frontend app.
Speed is everything for Vercel’s customers. They want to understand how their users are experiencing their application—all in real time. Does their site load quickly? Is it responsive? Is it visually stable? Accessing this data means that Vercel’s customers can draw insights and take action.
Vercel Analytics makes this process easy by collecting metrics from a user's device and calculating how well an application is performing in the wild. Analytics was originally built on AWS with Amazon Kinesis Data Firehose and Amazon Athena to ingest data and query application performance. For Joe Haddad, a Senior Software Engineer at Vercel, the challenge was that this data wasn’t in real time. Their setup wasn’t designed for analyzing huge volumes of data in realtime, at scale. For example, after enabling Analytics, it could take up to 30 minutes for data to be available to the user.
For Vercel’s customer success team, slow loading times for the Analytics dashboard meant more tickets raised, which in turn meant that greater resources were required to keep up with demand. For customers, it slowed down the speed at which they could understand their users' behavior, and iterate. And Speed is always a top business objective at Vercel.
When Vercel moved their analytics engine to Tinybird, everything changed. They no longer relied on their complex AWS set-up—Kinesis, Firehose, and Athena—and could instead use Tinybird to ingest data via HTTP and expose low-latency HTTP endpoints using SQL. They did all this with almost no guidance from the Tinybird team, while reducing their tooling costs.
Users now have a reactive dashboard experience, even when viewing data for long time spans. Data is available seconds after enabling Analytics, and users can iterate faster as they can immediately see updated metrics after a new production deployment.
Given the success of using Tinybird to make Vercel Analytics real time, the development team at Vercel saw the power of accessing other huge datasets in real time. They began experimenting with using Tinybird for other use cases, including real-time alerts, detecting and mitigating DDoS attacks, and adding network traffic information to Vercel Analytics. The team’s excited to continue expanding their use of Tinybird in the future.
Using a combination of Kafka and Tinybird's Data Source API. Some of the data is materialized on ingest, some of it is queried raw.
Vercel team builds SQL API endpoints in Tinybird they then consume from their application and some internal systems.
They use several Tinybird workspaces to organise their projects. Data can be shared amongst them and different developers tackle different projects