cockroachdb-in-comparison.md

title	summary	tags	toc	comparison
CockroachDB in Comparison	Learn how CockroachDB compares to other popular databases like PostgreSQL, Cassandra, MongoDB, Google Cloud Spanner, and more.	mongodb, mysql, dynamodb	false	true

This page shows you how the key features of CockroachDB stack up against other databases. Hover over the features for their intended meanings, and click CockroachDB answers to view related documentation.

	MySQL PostgreSQL Oracle AWS Aurora Cassandra MongoDB Spanner Yugabyte	MySQL PostgreSQL Oracle AWS Aurora Cassandra MongoDB Spanner Yugabyte	CockroachDB
Database horizontal scale	Manual Sharding Add on configuration Node based, automated read scale, limited write Node based, automated for both reads and writes	Manual Sharding Add on configuration Node based, automated read scale, limited write Node based, automated for both reads and writes	Node based, automated for both reads and writes
Database load balancing (internal)	Manual - not part of database None and full copies across regions Even distribution to optimize storage	Manual - not part of database None and full copies across regions Even distribution to optimize storage	Detailed options to optimize storage, compute and latency
Failover	Manual - not part of database Automated for reads, limited for writes to one region Automated for reads, limited guarantees for writes Fully automated for both reads and writes	Manual - not part of database Automated for reads, limited for writes to one region Automated for reads, limited guarantees for writes Fully automated for both reads and writes	Fully automated for both reads and writes
Automated repair and RPO(Recovery Point Objective)	Manual repair RPO ~1-60 mins Automated RPO ~1 -5 mins Manual & automated repair RPO <1 min "Automated repair RPO <10 sec"	Manual repair RPO ~1-60 mins Automated RPO ~1 -5 mins Manual & automated repair RPO <1 min "Automated repair RPO <10 sec"	Automated repair RPO = 0 sec
Distributed reads	Manual - asynchronous Yes	Manual - asynchronous Yes	Yes
Distributed transactions	No Lightweight transactions only Yes	No Lightweight transactions only Yes	Yes
Database isolation levels	Single region consistent default - Snapshot highest - Serializable Eventual consistent default - Read uncommitted highest - Snapshot read Eventual consistent - No transaction isolation guarantees Default - Snapshot highest - Serializable	Single region consistent default - Snapshot highest - Serializable Eventual consistent default - Read uncommitted highest - Snapshot read Eventual consistent - No transaction isolation guarantees Default - Snapshot highest - Serializable	Guaranteed consistent default - Serializable highest - Serializable
Potential data issues (default)	Phantom reads, non-repeatable reads, write skew Dirty reads, phantom reads, non-repeatable reads, write skew Dirty reads, phantom reads, non-repeatable reads, write conflicts None Phantom reads, non-repeatable reads	Phantom reads, non-repeatable reads, write skew Dirty reads, phantom reads, non-repeatable reads, write skew Dirty reads, phantom reads, non-repeatable reads, write conflicts None Phantom reads, non-repeatable reads	None
SQL	Yes No Yes - with limitations	Yes No Yes - with limitations	Yes - wire compatible with PostgreSQL
Database schema change	Yes Offline Online, Active and Dynamic	Yes Offline Online, Active and Dynamic	Online, Active and Dynamic
Cost based optimization	Yes No ? No	Yes No ? No	Yes
Data Geo-partitioning	No Yes, object level Yes No	No Yes, object level Yes No	Yes, row level
Upgrade method	Offline Online, rolling	Offline Online, rolling	Online, rolling
Multi-region	Yes - manual Yes, but not for writes Yes, for both reads and writes	Yes - manual Yes, but not for writes Yes, for both reads and writes	Yes for both reads and writes
Multi-cloud	No Yes	No Yes	Yes

* In DynamoDB, distributed transactions and ACID semantics across all data in the database, not just per row, requires an additional transaction library.