Serving Direct Messages At Twitter Scale

This is a past event

113 people went


I'm happy to announce Twitter will once again be hosting our meetup in June. If you're not familiar with the Twitter office in Boulder they are doing some really great work with some really big data. Chris Hayes and Mark Henderson from the Boulder Twitter team will be giving "lightning talks" on the Twitter infrastructure for handling Direct Messages. Come and hear some of the challenges and solutions the local Twitter team is working on.

6:00 – 6:45 - Socialize over food and drinks
6:45 – 7:00 - Welcome, opening remarks and announcements
7:00 – 8:30 – Chris Hayes, Mark Henderson & Shai Wilson - Lighting Talks
8:30 - 9:00 - Networking

About the event

Twitter’s Direct Messaging backend engineering team will deliver 3 lightning talks covering the DM engineering+product landscape.

1. When a server is overloaded by its clients, it’s a common reaction to begin rejecting traffic and expect the clients to back off in response. But what if you’re a client, and only a well-defined subset of your traffic is getting rejected? There are strategies for backing off only the subset, and Chris will explain one such strategy in use at Twitter today.

2. Twitter delivers a massive amount of Direct Messages sent every day. How do we reliably deliver every single message while maintaining consistent state across multiple clients and downstream systems? Mark will introduce the reliable delivery architecture behind Direct Messages.

3. Compared to pure infra projects, product projects have more stakeholders. There are engineers, researchers, the PM, executives, designers, lawyers, and many other parties who care about how the feature gets implemented. Everyone also has a different angle; however, working on a product feature requires quick decision making and process to reach a consensus.

About the Speakers

Chris Hayes

Chris Hayes has worked on myriad products and technologies over the years, from front-end to back-end and from media editing/flowchart tools to database abstraction layers. But, his passion has always been for human interactions and psychology, and he always tries to think about how the users will respond to a feature or use it unexpectedly.

Mark Henderson

Mark has been designing distributed backend systems since 2012. Before Twitter, he helped keep Amazon's fulfillment backend running, including integrations such as Kiva robotics and Amazon Fresh. He more recently worked on Google's payment processing systems. Outside the 9-5, he's an avid swing dancer in the Denver area.

Shai (pronounced shay) Wilson

Shai has been at Twitter for almost 3 years. She’s also made appearances at the United Nations, the Internet Archive, and Square. What really draws her into a particular project is how it aims to connect people. She sees user experience and personal detail as a potential to make a way bigger impact.