Event Sourcing

The foundational rule of Event Sourcing is that events are never deleted or modified. The event store acts as an append-only ledger. This guarantees that the system of record is a complete, conflict-free history of every state transition. Unlike CRUD-based systems where destructive updates overwrite history, this pattern preserves the true sequence of intent, enabling deep auditability and temporal querying.

The event stream is the single source of truth, not the current state. The current state of an entity is a left-fold derivation of its historical events. This means the database is a record of facts that happened, not a snapshot of the world right now. If a bug corrupts a projection, the system can discard the derived state and rebuild it deterministically from the immutable event log.

Because the full history is stored, you can reconstruct the state of any entity at any point in time. This enables time travel queries to answer complex business questions like 'What was the shopping cart value before the discount was applied?' This is critical for retroactive debugging, financial auditing, and compliance reporting where point-in-time accuracy is mandatory.

Event Sourcing is often paired with Command Query Responsibility Segregation (CQRS) . The write model processes commands and appends events, while the read model subscribes to the event stream to build optimized projections. This separation allows the write side to remain simple and fast, while the read side can be scaled independently and denormalized for specific query patterns without affecting the integrity of the source events.

Every state change is captured as a distinct event with metadata including the timestamp, user intent, and causation identifier. This creates an inherent, tamper-proof audit log without requiring separate logging infrastructure. For regulated industries, this answers not just 'who changed what,' but 'why was it changed,' providing a forensic-level view of business process execution.

The append-only log naturally functions as a reliable message bus. Other services and projections can subscribe to the event stream to react to business occurrences in real-time. This transforms the event store into a durable, replayable backbone for a distributed, event-driven architecture, ensuring downstream consumers never miss a critical business signal even if they are temporarily offline.

Modern banking cores and payment processors use event sourcing to maintain an immutable audit trail of every transaction. Instead of storing a mutable 'current balance,' the system persists a sequence of FundsDeposited, FundsWithdrawn, and FeeApplied events.

• Full Auditability: Regulators can replay the event log to reconstruct the account state at any point in time.
• Correction, Not Deletion: Errors are fixed by appending a compensating CorrectionIssued event, preserving the original mistake for forensic analysis.
• Real-world example: The Ledger system at a major European neobank uses this pattern to achieve 100% traceability across millions of accounts.

Electronic Health Record (EHR) systems leverage event sourcing to comply with strict regulations like HIPAA and GDPR. Every interaction with a patient record—from a nurse administering medication to a specialist updating a diagnosis—is captured as an immutable event.

• Temporal Queries: Clinicians can view a patient's condition as it was on a specific date, not just the latest snapshot.
• Tamper-Proof Compliance: The append-only log proves that records have not been altered, only appended to, satisfying legal chain-of-custody requirements.
• Example: A leading hospital network uses event-sourced patient journals to reconstruct the exact sequence of clinical decisions during critical incident investigations.

Global logistics platforms track millions of packages through a complex web of handlers. Event sourcing models the journey as a series of state transitions: PackageReceived, CustomsCleared, OutForDelivery, Delivered.

• Full Provenance: Stakeholders can query the exact location and custodian of a package at any historical moment.
• Anomaly Detection: Replaying the event stream allows machine learning models to identify bottlenecks and predict delays based on deviation from ideal event sequences.
• Real-world example: A major container shipping line uses an event-sourced Digital Twin to provide real-time, auditable visibility to every party in the supply chain.

Complex order lifecycles cannot be modeled as a simple state machine with a mutable status field. Event sourcing captures every nuance: OrderPlaced, PaymentAuthorized, ItemBackordered, ShipmentSplit, PartialRefundIssued.

• Rebuild State: The current order status is a left-fold of all historical events, eliminating inconsistent states.
• Debugging & Analytics: Customer service teams can replay the exact sequence of events to understand why an order ended up in a specific state, reducing resolution time.
• Example: A high-volume flash-sale platform uses event sourcing to handle complex partial fulfillment and cancellation logic without losing data integrity.

In industrial IoT, sensor data is a continuous stream of immutable facts. Event sourcing stores raw telemetry—TemperatureExceeded, VibrationSpikeDetected, MaintenanceScheduled—as the system of record.

• Time Travel Debugging: Engineers can replay sensor data from the moment before a turbine failure to understand the cascade of events that led to the malfunction.
• Dual-Timeline Architecture: The event log provides the 'as-it-happened' timeline, while projections build the 'corrected-for-clock-drift' timeline for analytics.
• Real-world example: A wind farm operator uses event sourcing to store petabytes of immutable sensor data, enabling predictive maintenance models to train on exact historical sequences.

Contract negotiation involves a delicate sequence of proposals, counter-proposals, and amendments. Event sourcing treats the contract not as a final document, but as a stream of ClauseProposed, ClauseAccepted, and ContractExecuted events.

• Non-Repudiation: Every party's actions are cryptographically signed and appended to the log, creating a definitive record of intent.
• Branching & Merging: The event log can model parallel negotiation tracks (redlines) that are eventually merged, preserving the full history of who changed what and when.
• Example: A legal tech startup uses event sourcing to provide an immutable negotiation transcript that holds up in court as a record of the parties' meeting of the minds.