Building Systems That Go Fast and Fail Safely
Talk 1: Automated JDBC query caching with Valkey
Modern Java applications built on relational databases often face a common challenge: as application traffic grows, read queries become the dominant source of database load. This is especially true for frequently accessed but infrequently updated data. Often times those queries require joins and aggregations across large data sets to fetch just a few rows of results. This can be very resource constraint and may even lead to performance issues. AWS Advanced JDBC wrapper presents a solution to this challenge by allowing caching query results from expensive and repetitive read queries to an in-memory cache like Valkey. You can dramatically reduce database load and improve query latency while maintaining or even improving application performance.
Talk 2: Hermetically Sealed Box of Pain
Most programmers encounter property-based testing as “QuickCheck, but for pure functions.”
But what happens when the thing you want to test isn’t a function? What if it’s a distributed database, a consensus protocol, or an entire network of stateful services yelling at each other over TCP? This talk starts with classic property testing and scales the idea all the way up to Deterministic Simulation Testing (DST): controlling time, scheduling, crashes, retries, and randomness so entire systems become testable and searchable.
We’ll look at techniques and systems from QuickCheck, FoundationDB, TigerBeetle’s VOPR, Antithesis, and others, and explore the blurry line between testing, fuzzing, and lightweight formal methods.
This is a deep dive into how modern reliability testing actually works, aimed at people who enjoy distributed systems, weird bugs, and making computers fight for their lives in a hermetically sealed box of pain.
About Speakers
Qu Chen is a Senior Software Development Engineer with more than 10 years of technical expertise in AWS Cloud Technology and in-memory database services, responsible in building and maintaining Amazon ElastiCache for Valkey/Memcached and Amazon MemoryDB. As an active contributor in the Valkey open-source community, Qu has given talks and published posts on building low-latency applications with caching.
Carl Sverre is a systems engineer and open-source builder passionate about the foundations of reliable software. Currently at Antithesis, he works on automated testing at scale — helping teams catch hard-to-reproduce bugs in complex distributed systems. He is the creator of SQLSync, an offline-first multiplayer SQLite platform, and Graft, a transactional storage engine optimized for lazy partial replication. Before Antithesis, Carl spent nearly ten years at MemSQL (now SingleStore), rising to Senior Director, and later served as Entrepreneur in Residence at Amplify Partners.