November Edition @ codecentric: Taming CRDTs and Tail Latency

We're back on November 11th for another evening of deep database technical discussions! This time, we're hosted at codecentric AG, featuring talks on tail latency optimization and CRDT implementations over constrained networks.

As always, there will be plenty of time for networking, discussions, and pizza 🍕.

Event Details
📍 Venue: codecentric AG, August-Everding-Straße 20, München
📅 Date & Time: Tuesday, November 11th, doors open at 18:30

***

The Tale of Taming TigerBeetle's Tail Latency - Tobias Ziegler, TigerBeetle

In this talk, we explore how TigerBeetle's tail latency was reduced through algorithm engineering. Unlike traditional algorithm analysis, algorithm engineering focuses on bridging theory and practice - designing algorithms that not only look good on paper but also run efficiently on modern superscalar CPUs. We'll walk through a mix of high-level design choices and low-level micro-optimizations, showing how both work together to reduce tail latency in practice.

Speaker Bio: Tobias works as a software engineer at TigerBeetle, where he focuses on improving system performance. Before joining TigerBeetle, he was a Postdoctoral Researcher at the Chair of Decentralized Information Systems and Data Management at the Technical University of Munich, focusing on cost-efficient, cloud-native data systems. He earned his PhD in Computer Science from TU Darmstadt in 2023, where his research centered on distributed databases and modern network technologies such as RDMA. In 2021, he won the 2021 SIGMOD Best Paper Award.

***

CRDTs over Painful Networks - Jon Gjengset & Oliver Wangler, Helsing

The best known uses of CRDTs are for collaborative text editing. In that domain, the CRDT semantics help reconcile concurrent edits by multiple online users, as well as users who go offline, make edits, and come back online again later. But the power of CRDTs extends far beyond this binary-online-or-offline use-case.

At Helsing, we're using CRDTs to share situational awareness over networks that are often severely degraded; through long distances, radio silence, moving nodes, and jamming. And this is where CRDTs truly shine, giving high availability and partition tolerance while ensuring eventual consistency when connectivity is available. Through the use of delta-state, data schemas, custom serialisation protocols, and more, we can operate even over network bearers with severely constrained bandwidth. In this talk, we'll go through some of the mechanisms that make this possible, the DSON crate we've open-sourced as part of the work, and some observations about interesting open questions that remain.

Speaker Bios: Jon Gjengset is a Principal Engineer at Helsing, author of Rust for Rustaceans, and general Rust educator. Oliver Wangler is a hacker by day and by night, focused on coding systems and seeing them do what they were intended to do in production.

***

Agenda
🔹 18:30: Doors Open
🔹 18:40: Welcome
🔹 18:45: Talk #1: Tobias Ziegler (TigerBeetle)
🔹 19:30: Pizza & Networking 🍕
🔹 20:00: Talk #2: Jon Gjengset & Oliver Wangler (Helsing)

This event will provide deep technical insights into performance optimization at the CPU level, distributed consistency models, and practical CRDT implementations in production systems. Whether you're working on database internals, distributed systems, or high-performance computing, expect an evening of practical implementation details and battle-tested approaches.

See you there! 🚀