https://gorkemsari.github.io/distributed-systems/en/whats-new/ New and rewritten pages of the Distributed Systems Handbook. en Wed, 11 Nov 2026 06:00:00 +0000 https://gorkemsari.github.io/distributed-systems/en/data-management/replication/leaderless-replication/ data-management/replication/leaderless-replication#2026-11-11 Wed, 11 Nov 2026 06:00:00 +0000 How Dynamo-style stores send every write to N replicas and wait for W, read R, and why overlapping quorums still fall short of linearizability. https://gorkemsari.github.io/distributed-systems/en/data-management/replication/multi-leader-replication/ data-management/replication/multi-leader-replication#2026-11-10 Tue, 10 Nov 2026 06:00:00 +0000 How several leaders accept writes and replicate to each other, why concurrent writes conflict, and how detection and resolution must converge in production. https://gorkemsari.github.io/distributed-systems/en/data-management/replication/replication-lag/ data-management/replication/replication-lag#2026-11-09 Mon, 09 Nov 2026 06:00:00 +0000 Why reads from lagging followers break read-your-writes, monotonic and prefix order, and how time windows, LSN/GTID tokens and waits restore them. https://gorkemsari.github.io/distributed-systems/en/data-management/replication/failover-leader-election/ data-management/replication/failover-leader-election#2026-11-08 Sun, 08 Nov 2026 06:00:00 +0000 How a replicated database detects a failed leader, elects one successor without losing acknowledged writes, and fences the old one, with real tools. https://gorkemsari.github.io/distributed-systems/en/data-management/replication/leader-based-replication/ data-management/replication/leader-based-replication#2026-11-07 Sat, 07 Nov 2026 06:00:00 +0000 How a leader ships an ordered change log to followers, what async, semi-sync and sync acks guarantee, and how PostgreSQL, MySQL and MongoDB do it. https://gorkemsari.github.io/distributed-systems/en/data-management/data-partitioning/ data-management/data-partitioning#2026-11-06 Fri, 06 Nov 2026 06:00:00 +0000 How a dataset too large for one node is split so each key has one owner, why placement trades locality for load spread, and how to read this chapter. https://gorkemsari.github.io/distributed-systems/en/data-management/data-partitioning/secondary-indexes/ data-management/data-partitioning/secondary-indexes#2026-11-06 Fri, 06 Nov 2026 06:00:00 +0000 How secondary indexes work in partitioned stores: local indexes that scatter reads, global ones that fan out writes, how each stays correct, and failures. https://gorkemsari.github.io/distributed-systems/en/data-management/data-partitioning/colocated-joins/ data-management/data-partitioning/colocated-joins#2026-11-06 Fri, 06 Nov 2026 06:00:00 +0000 How one placement key for related tables makes distributed joins and transactions local, how Citus, Vitess and Spanner do it, and how colocation fails. https://gorkemsari.github.io/distributed-systems/en/data-management/data-partitioning/dynamic-rebalancing/ data-management/data-partitioning/dynamic-rebalancing#2026-11-06 Fri, 06 Nov 2026 06:00:00 +0000 How partitioned stores move data between nodes under live traffic: copy, catch-up and fenced cutover, with Kafka, MongoDB and Vitess, and how moves fail. https://gorkemsari.github.io/distributed-systems/en/data-management/data-partitioning/consistent-hashing/ data-management/data-partitioning/consistent-hashing#2026-11-06 Fri, 06 Nov 2026 06:00:00 +0000 How a hash ring moves only a minimal share of keys on membership change, why virtual nodes fix skew, and how vnode count trades balance for repair cost. https://gorkemsari.github.io/distributed-systems/en/data-management/data-partitioning/skew-hot-spots/ data-management/data-partitioning/skew-hot-spots#2026-11-06 Fri, 06 Nov 2026 06:00:00 +0000 Why a partitioned system is limited by its hottest partition, where skew comes from, how to find heavy keys, and which techniques remove hot spots. https://gorkemsari.github.io/distributed-systems/en/data-management/data-partitioning/hash-partitioning/ data-management/data-partitioning/hash-partitioning#2026-11-06 Fri, 06 Nov 2026 06:00:00 +0000 How hash partitioning maps keys via modulo, fixed slots or hash ranges, how uniform the result really is, and how mismatched hashes and resizing break it. https://gorkemsari.github.io/distributed-systems/en/data-management/data-partitioning/key-range-partitioning/ data-management/data-partitioning/key-range-partitioning#2026-11-06 Fri, 06 Nov 2026 06:00:00 +0000 How range-partitioned stores split sorted keys, route and split ranges under live traffic, and why the order that makes scans cheap creates hot spots. https://gorkemsari.github.io/distributed-systems/en/communication/ communication#2026-11-06 Fri, 06 Nov 2026 06:00:00 +0000 How services exchange requests and events over an unreliable network, and how this part's chapters on calls, messaging, contracts and balancing fit. https://gorkemsari.github.io/distributed-systems/en/communication/load-balancing/ communication/load-balancing#2026-11-06 Fri, 06 Nov 2026 06:00:00 +0000 How interchangeable replicas are made to act as one service, why each balancing decision trades information quality for cost, and how to read this chapter. https://gorkemsari.github.io/distributed-systems/en/communication/load-balancing/health-aware-routing/ communication/load-balancing/health-aware-routing#2026-11-06 Fri, 06 Nov 2026 06:00:00 +0000 How load balancers combine health checks with outlier detection, why ejection caps and panic thresholds matter, and how to drain a backend cleanly. https://gorkemsari.github.io/distributed-systems/en/communication/load-balancing/gslb-anycast/ communication/load-balancing/gslb-anycast#2026-11-06 Fri, 06 Nov 2026 06:00:00 +0000 How DNS-based global load balancing and BGP anycast steer users across regions, how fast each reacts to failure, and why production edges combine the two. https://gorkemsari.github.io/distributed-systems/en/communication/load-balancing/client-vs-server-side/ communication/load-balancing/client-vs-server-side#2026-11-06 Fri, 06 Nov 2026 06:00:00 +0000 Whether load balancing runs in a proxy tier or inside every client, and how that choice shapes discovery, failure detection and connection counts. https://gorkemsari.github.io/distributed-systems/en/communication/load-balancing/algorithms/ communication/load-balancing/algorithms#2026-11-05 Thu, 05 Nov 2026 06:00:00 +0000 How round robin, least connections, weighted and hash-based balancing pick a backend, what each actually equalizes, and how each fails in production. https://gorkemsari.github.io/distributed-systems/en/communication/load-balancing/l4-vs-l7/ communication/load-balancing/l4-vs-l7#2026-11-05 Thu, 05 Nov 2026 06:00:00 +0000 How layer 4 balancers pin flows from packet headers, how layer 7 proxies terminate and re-issue each request, and what each costs in CPU and failures. https://gorkemsari.github.io/distributed-systems/en/communication/api-gateway/ communication/api-gateway#2026-11-05 Thu, 05 Nov 2026 06:00:00 +0000 How request-response APIs are exposed safely across teams, why edge policies are cheap but not authoritative, and how to read this chapter's six topics. https://gorkemsari.github.io/distributed-systems/en/communication/api-gateway/throttling-rate-limiting/ communication/api-gateway/throttling-rate-limiting#2026-11-05 Thu, 05 Nov 2026 06:00:00 +0000 How API gateways enforce per-caller rate limits across many instances: local vs. global counters, the 429 contract, product differences, and failures. https://gorkemsari.github.io/distributed-systems/en/communication/api-gateway/idempotency/ communication/api-gateway/idempotency#2026-11-05 Thu, 05 Nov 2026 06:00:00 +0000 What safe and idempotent HTTP methods allow, how idempotency keys make POST retries safe, and how lease expiry and replica reads bring duplicates back. https://gorkemsari.github.io/distributed-systems/en/communication/api-gateway/contract-testing/ communication/api-gateway/contract-testing#2026-11-05 Thu, 05 Nov 2026 06:00:00 +0000 How consumer-driven contracts with Pact capture real consumer expectations, verify them against providers, and become a deployment gate, plus how they fail. https://gorkemsari.github.io/distributed-systems/en/communication/api-gateway/openapi-swagger/ communication/api-gateway/openapi-swagger#2026-11-05 Thu, 05 Nov 2026 06:00:00 +0000 How OpenAPI models an HTTP API, how a contract-first pipeline lints, diffs, mocks and validates it, and what that pipeline cannot prove in production. https://gorkemsari.github.io/distributed-systems/en/communication/api-gateway/bff-pattern/ communication/api-gateway/bff-pattern#2026-11-05 Thu, 05 Nov 2026 06:00:00 +0000 How a Backend for Frontend composes data for one client, why it must stay free of domain rules, and how its fan-out latency and failures add up. https://gorkemsari.github.io/distributed-systems/en/communication/api-gateway/api-gateway-pattern/ communication/api-gateway/api-gateway-pattern#2026-11-05 Thu, 05 Nov 2026 06:00:00 +0000 How an API gateway routes and authenticates requests at the edge, which invariants keep edge authentication sound, and how a shared entry layer fails. https://gorkemsari.github.io/distributed-systems/en/communication/async-messaging/ communication/async-messaging#2026-11-05 Thu, 05 Nov 2026 06:00:00 +0000 Why a durable broker trades immediate, visible failure for delayed, invisible failure, and how to read this chapter's eight messaging topics. https://gorkemsari.github.io/distributed-systems/en/communication/async-messaging/dlq-poison-messages/ communication/async-messaging/dlq-poison-messages#2026-11-05 Thu, 05 Nov 2026 06:00:00 +0000 How consumers tell poison messages from transient failures, how bounded retries and dead-letter queues prevent stalls, and how to redrive a DLQ safely. https://gorkemsari.github.io/distributed-systems/en/communication/async-messaging/schema-evolution-registry/ communication/async-messaging/schema-evolution-registry#2026-11-05 Thu, 05 Nov 2026 06:00:00 +0000 How message schemas evolve without breaking readers, what each compatibility mode guarantees, and how Avro, Protobuf and a schema registry enforce it. https://gorkemsari.github.io/distributed-systems/en/communication/async-messaging/asyncapi/ communication/async-messaging/asyncapi#2026-11-05 Thu, 05 Nov 2026 06:00:00 +0000 How an AsyncAPI document describes an application's channels and messages, what it can guarantee across independent services, and how to keep it accurate. https://gorkemsari.github.io/distributed-systems/en/communication/async-messaging/delivery-semantics/ communication/async-messaging/delivery-semantics#2026-11-05 Thu, 05 Nov 2026 06:00:00 +0000 Why lost messages and lost acks force a choice between loss and duplicates, and how deduplication turns at-least-once delivery into exactly-once effects. https://gorkemsari.github.io/distributed-systems/en/communication/async-messaging/rabbitmq/ communication/async-messaging/rabbitmq#2026-11-05 Thu, 05 Nov 2026 06:00:00 +0000 How RabbitMQ routes messages through exchanges and bindings, how each exchange type matches, and what happens to unroutable messages. https://gorkemsari.github.io/distributed-systems/en/communication/async-messaging/pulsar-vs-kafka/ communication/async-messaging/pulsar-vs-kafka#2026-11-05 Thu, 05 Nov 2026 06:00:00 +0000 How Pulsar separates stateless brokers from BookKeeper storage, how that differs from Kafka's broker-owned logs, and what each costs to operate. https://gorkemsari.github.io/distributed-systems/en/communication/async-messaging/kafka-architecture/ communication/async-messaging/kafka-architecture#2026-11-05 Thu, 05 Nov 2026 06:00:00 +0000 How Kafka builds replicated, ordered partitions, how the ISR and high watermark decide which acknowledged writes survive, and what offsets mean. https://gorkemsari.github.io/distributed-systems/en/communication/async-messaging/message-queue-vs-event-streaming/ communication/async-messaging/message-queue-vs-event-streaming#2026-11-05 Thu, 05 Nov 2026 06:00:00 +0000 How queues delete messages on acknowledgement while event logs keep an ordered history read by offset, and how that decides ordering, replay and fan-out. https://gorkemsari.github.io/distributed-systems/en/communication/synchronous-protocols/ communication/synchronous-protocols#2026-11-05 Thu, 05 Nov 2026 06:00:00 +0000 Why a caller waiting on a remote answer inherits the callee's latency and failures, how the protocols compare, and how to read this chapter's seven topics. https://gorkemsari.github.io/distributed-systems/en/communication/synchronous-protocols/api-versioning/ communication/synchronous-protocols/api-versioning#2026-11-05 Thu, 05 Nov 2026 06:00:00 +0000 What makes an API change breaking, where version identifiers can live and how that affects routing and caching, and how to retire old versions safely. https://gorkemsari.github.io/distributed-systems/en/communication/synchronous-protocols/websocket-sse/ communication/synchronous-protocols/websocket-sse#2026-11-04 Wed, 04 Nov 2026 06:00:00 +0000 How WebSocket handshakes and framing work, how Server-Sent Events stream and resume, and how to run and drain millions of long-lived connections. https://gorkemsari.github.io/distributed-systems/en/communication/synchronous-protocols/graphql/ communication/synchronous-protocols/graphql#2026-11-04 Wed, 04 Nov 2026 06:00:00 +0000 How GraphQL executes client-shaped queries, why resolvers cause N+1 fan-out and how batching fixes it, how to bound query cost, and how federation works. https://gorkemsari.github.io/distributed-systems/en/communication/synchronous-protocols/grpc-streaming/ communication/synchronous-protocols/grpc-streaming#2026-11-04 Wed, 04 Nov 2026 06:00:00 +0000 How gRPC's four call types map onto HTTP/2 streams, how flow control becomes backpressure, and why long-lived streams break retries and balancing. https://gorkemsari.github.io/distributed-systems/en/communication/synchronous-protocols/grpc-protocol-buffers/ communication/synchronous-protocols/grpc-protocol-buffers#2026-11-04 Wed, 04 Nov 2026 06:00:00 +0000 How Protocol Buffers encodes messages and stays compatible, how gRPC maps calls onto HTTP/2 with deadlines and cancellation, and how the stack fails. https://gorkemsari.github.io/distributed-systems/en/communication/synchronous-protocols/http2-http3-quic/ communication/synchronous-protocols/http2-http3-quic#2026-11-04 Wed, 04 Nov 2026 06:00:00 +0000 How HTTP/2 multiplexes streams over one TCP connection, why head-of-line blocking survives it, and how HTTP/3 over QUIC removes it, at what cost. https://gorkemsari.github.io/distributed-systems/en/communication/synchronous-protocols/rest/ communication/synchronous-protocols/rest#2026-11-04 Wed, 04 Nov 2026 06:00:00 +0000 What REST constraints buy and cost, how HTTP methods, status codes and cache headers form a contract, and how lost updates and duplicates break it. https://gorkemsari.github.io/distributed-systems/en/foundations/time-and-ordering/distributed-id-generation/ foundations/time-and-ordering/distributed-id-generation#2026-11-04 Wed, 04 Nov 2026 06:00:00 +0000 Where uniqueness comes from in Snowflake, UUIDv7 and ULID, how they handle bursts and backward clocks, how ordering affects indexes, and how they fail. https://gorkemsari.github.io/distributed-systems/en/foundations/time-and-ordering/truetime-spanner/ foundations/time-and-ordering/truetime-spanner#2026-11-04 Wed, 04 Nov 2026 06:00:00 +0000 How TrueTime turns clock error into a bounded interval, how Spanner's commit wait yields external consistency, what it costs, and what a wrong bound breaks. https://gorkemsari.github.io/distributed-systems/en/foundations/time-and-ordering/hybrid-logical-clocks/ foundations/time-and-ordering/hybrid-logical-clocks#2026-11-04 Wed, 04 Nov 2026 06:00:00 +0000 How hybrid logical clocks keep causality while staying close to wall time, what they enable for MVCC snapshots, and how one bad clock spreads. https://gorkemsari.github.io/distributed-systems/en/foundations/time-and-ordering/vector-clocks/ foundations/time-and-ordering/vector-clocks#2026-11-04 Wed, 04 Nov 2026 06:00:00 +0000 How vector clocks capture happened-before exactly, how version vectors apply them to replicated data, what the metadata costs, and how they break. https://gorkemsari.github.io/distributed-systems/en/foundations/time-and-ordering/lamport-timestamps/ foundations/time-and-ordering/lamport-timestamps#2026-11-04 Wed, 04 Nov 2026 06:00:00 +0000 How Lamport timestamps never invert a causal pair, how a tie-break yields a total order, how Raft terms and Paxos ballots use it, and how implementations fail. https://gorkemsari.github.io/distributed-systems/en/foundations/time-and-ordering/happened-before/ foundations/time-and-ordering/happened-before#2026-11-04 Wed, 04 Nov 2026 06:00:00 +0000 Lamport's happened-before relation as the order a distributed system can actually observe, why it defines causality, and what it makes possible.