Anonim / 5 days, 5 hours ago | Download | Plain text | Reply |

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
# Event Sourcing and CQRS: When to Mention Them in an Interview

## Event sourcing

Instead of storing the current state of an entity, store the sequence of **events** that happened to it: `AccountOpened`, `MoneyDeposited(50)`, `MoneyWithdrawn(20)`. Current state is computed by replaying events (balance = 30).

Benefits:

- A complete audit log for free, which is why finance and ledger systems like it.
- You can rebuild state as of any point in time, or build new views from old history.
- Debugging: you can see exactly how an entity got into a bad state.

Costs:

- Replaying long histories is slow, so you store periodic **snapshots** and replay only newer events.
- Changing event formats needs versioning ("upcasting" old events).
- Queries like "all accounts with balance over 1000" need a separate read model.

## CQRS

Command Query Responsibility Segregation separates the **write model** (validates commands and appends events) from one or more **read models** (projections optimized for queries: a SQL table for lists, a search index for text search, a cache for dashboards). Projections are updated asynchronously from the event stream.

The read side is eventually consistent, so the UI must tolerate a short delay after a write (or read its own writes from the write side).

## When to bring it up

Good fits: ledgers, order lifecycles, collaborative tools with history, systems that need audit trails. Poor fits: simple CRUD apps, where it adds a lot of complexity for little gain. Saying that out loud shows judgment.

Disclosure: I'm part of the team at Linkjob AI (https://www.linkjob.ai/), an AI interview copilot.