Tag: design-distributed
All the articles with the tag "design-distributed".
-
System Design - How Hedged Requests Reduce Tail Latency
Follow a slow read through a hedged RPC, from the delayed second attempt to cancellation, deadlines, and the cost of duplicate work.
-
System Design - How Idempotency Keys Make Retries Safe
Follow a payment request whose response is lost: how a stable request key, atomic result recording, and careful retry boundaries prevent a second charge.
-
System Design - How Sagas Coordinate Distributed Transactions
Follow an order through inventory reservation, payment, and compensation to understand what a saga guarantees, what it cannot undo, and how to make retries safe.
-
System Design - How the Transactional Outbox Works
Why saving a database row and publishing an event can fail, and how a transactional outbox connects PostgreSQL, Debezium, Kafka, and an idempotent consumer without pretending delivery happens exactly once.
-
System Design - How Distributed Stream Processing Works (Apache Flink)
How distributed stream processing works: dataflow graphs, event time vs processing time, windowing, watermarks, state management, checkpointing, and exactly-once semantics — with source code walkthrough from Apache Flink.
-
System Design - How Distributed File Systems Work (GFS/HDFS)
How distributed file systems work: the architecture of Google File System and Hadoop HDFS, chunk-based storage, metadata management, replication, read/write flows, and fault tolerance — with source code references.