Explains idempotency keys in distributed systems, comparing UUIDs and monotonic sequences for duplicate detection and exactly-once processing.
Explores building a basic Durable Execution engine using SQLite and Java to reliably persist and resume multi-step workflows, like those in agentic systems.
An engineer argues that software development is a learning process, not an assembly line, and explains how to use LLMs as brainstorming partners.
Explains the concept of a 'synchrony budget' for designing distributed systems, advocating for asynchronous communication to improve performance and availability.
Explores KIP-932, a proposal to add queue semantics and share groups to Apache Kafka for improved message processing.
Explains how to implement distributed leader election using Amazon S3's new conditional write feature for creating a lock file.
An introduction to NATS, an open-source technology for connecting applications and microservices over a network.
Explores challenges of running Kafka Connect on Kubernetes and proposes a vision for a more Kubernetes-native architecture.
An analysis of different ID types in distributed systems, comparing transparent vs. opaque identifiers and their tradeoffs.
Demystifying eventual consistency in DynamoDB, explaining its trade-offs and how to handle it effectively in applications.
A guide to understanding the CAP theorem in distributed systems, explaining when it applies and its implications for consistency and availability.
Explores the three different meanings of 'consistency' in databases: from the CAP Theorem, ACID properties, and database consistency models.
An analysis of Twitter's most severe cache-related incidents from 2012-2022, exploring patterns and knowledge loss.
A discussion of common pitfalls in measuring tail latency metrics in distributed systems, using examples from Twitter's infrastructure.
Explores strategies for handling Kafka message delivery failures in applications, discussing trade-offs between consistency and availability.
A recap of 2021 conference talks on Debezium and Change Data Capture (CDC), exploring patterns and integrations with tools like Kafka and Infinispan.
The author argues that measurement and data analysis are undervalued but high-ROI activities, using the impact of Jepsen tests as a key example.
An overview of Kafka's new KRaft mode, which removes the ZooKeeper dependency for metadata management and controller election.
A developer explains why they stopped using microservices for their solo side-projects, citing the complexity for a single person.
Argues that the PIE theorem (Pattern Flexibility, Efficiency, Infinite Scale) is more practical for database selection than the popular CAP theorem.
