Event-Sourcing Technology Landscape — Event Sourcing & CQRS

You can event-source with plain tables, but frameworks help with aggregates, projections and subscriptions: Axon Framework (Java), Marten and Eventuous (.NET), EventStoreDB/KurrentDB clients (many languages), Commanded (Elixir), eventsourcing (Python library), Emmett (TypeScript), and DynamoDB/Cosmos DB-based designs. Choose based on stack, operational appetite and whether you need projections and subscriptions out of the box.

Libraries and stores

Stack	Options	Store
Java/Kotlin	Axon Framework, EventStoreDB client	Axon Server, RDBMS, EventStoreDB
.NET	Marten, Eventuous	PostgreSQL, EventStoreDB
Python	eventsourcing library	PostgreSQL, SQLite, EventStoreDB
TypeScript/Node	Emmett, EventStoreDB client	PostgreSQL, EventStoreDB, MongoDB
Elixir	Commanded	PostgreSQL (EventStore), EventStoreDB
Cloud-native	DynamoDB streams pattern	DynamoDB + Streams

# DynamoDB event store pattern
PK = "account#42", SK = "v#000005" (stream version), conditional put: attribute_not_exists(SK)
DynamoDB Streams → Lambda projectors → read models

Common pitfalls

Adopting a heavyweight framework for one small aggregate.
Choosing a store without subscription/catch-up support.
Vendor lock-in without an export strategy.

Key points

Frameworks reduce boilerplate
Postgres-based options are pragmatic
Check subscriptions and tooling

Further reading: Python eventsourcing library

Practice questions

How can DynamoDB enforce optimistic concurrency for event streams?

Answer

Use the stream version as the sort key with a conditional put (attribute_not_exists), so two writers can't append the same version.

Why is PostgreSQL a popular event store choice?

Answer

It's familiar and transactional, supports JSONB payloads, unique constraints for concurrency, and libraries like Marten build event-store features on top.