Backend Development › Transactions & Concurrency Control

ACID

Atomicity, Consistency, Isolation, Durability.

Backend Engineer · Mid-level · core Data Engineer · Mid-level

Also known as: ACID properties, ACID transactions, ACID compliance, atomicity consistency isolation durability

ACID names the four guarantees that a reliable database transaction provides. They’re why you can trust a relational database with money.

Letter	Property	Meaning
A	Atomicity	A transaction is all or nothing. If any step fails, everything is rolled back (atomicity)
C	Consistency	A transaction takes the database from one valid state to another: constraints (keys, uniqueness, checks) always hold
I	Isolation	Concurrent transactions don’t see each other’s half-finished work. It behaves as if they ran one at a time (to a degree you choose, see isolation levels)
D	Durability	Once committed, the change survives crashes and power loss (the database writes to a log on disk before saying “done”, see WAL)

BEGIN;
UPDATE accounts SET balance = balance - 100 WHERE id = 1;
UPDATE accounts SET balance = balance + 100 WHERE id = 2;
COMMIT;          -- both changes happen, or neither (A); the total stays valid (C); no one sees only half (I); it survives a crash (D)

Nuances worth knowing

Consistency in ACID is about the database’s own rules. It isn’t the same as “consistency” in distributed systems (all replicas agree), which is a different concept.
Isolation is a spectrum. Full serializable isolation is the strongest and slowest. Most databases default to weaker levels (read committed, repeatable read), which allow some anomalies such as non-repeatable reads, phantom reads and lost updates (dirty reads, lost updates, write skew). You need to know what yours guarantees.
Durability depends on configuration. Settings that skip flushing to disk trade safety for speed.
ACID covers one database. Across several services or databases, you need other approaches (distributed transactions, sagas).

ACID and NoSQL

Traditional relational databases are ACID. Many NoSQL systems relax some guarantees for scale or availability. Some offer ACID within a single record or partition, and many modern ones also support multi-document transactions. Check the specific guarantees instead of assuming by category (SQL vs NoSQL).

If someone says “use a database with ACID”, they mean: don’t lose or corrupt my data when things fail or run at the same time.