NewSQL Database
Combines the horizontal scalability of NoSQL with the ACID guarantees and SQL interface of traditional relational databases. Optimized for globally distributed applications that refuse to compromise between consistency and scale.
Character
The overachiever who insists on having it all: the scalability of NoSQL and the consistency of SQL, no compromises accepted. It's the answer to the question 'Why can't we just scale our PostgreSQL globally?' Born from the frustration of choosing between consistency and scale.
When to Use
- Global applications requiring strong consistency across regions
- Migrating from PostgreSQL or MySQL that has outgrown single-node scaling
- Financial systems needing distributed ACID transactions
- Multi-region SaaS platforms with regulatory data residency requirements
Avoid When
- Single-region deployment where a traditional RDBMS suffices
- Latency-sensitive workloads that cannot tolerate consensus overhead
- Budget constraints make managed distributed databases impractical
Dimension Analysis
↑ Scalability
Designed for horizontal scaling with automatic sharding and rebalancing. CockroachDB and Spanner distribute data globally while maintaining strong consistency, with near-linear write scaling across regions.
⚡ Performance
Distributed consensus (Raft/Paxos) adds latency compared to single-node RDBMS. Cross-region transactions can take 100 to 300ms. Local reads are fast, but globally consistent writes have inherent network overhead.
⚓ Reliability
Serializable isolation and distributed ACID transactions are the defining feature. Google Spanner achieves 99.999% availability, and CockroachDB survives entire region failures without data loss through Raft-based replication.
⚙ Operational Simplicity
Cluster topology, range rebalancing, and multi-region configuration add significant operational complexity. While easier than manual sharding of PostgreSQL, it still requires understanding distributed systems concepts.
⯑ Query Flexibility
Full SQL support including joins, indexes, transactions, and foreign keys. CockroachDB is wire-compatible with PostgreSQL. Some limitations exist on distributed joins and certain advanced SQL features.
⧉ Schema Flexibility
Inherits relational schema rigidity with ALTER TABLE, migrations, and foreign key constraints. Schema changes on large distributed datasets require careful online DDL strategies to avoid downtime.
★ Ecosystem Maturity
Younger than traditional RDBMS (CockroachDB launched in 2015, TiDB in 2015). Growing adoption in cloud-native enterprises, but the tooling ecosystem and community are smaller than PostgreSQL or MySQL.
↗ Learning Curve
SQL familiarity makes the transition approachable, but understanding distributed transaction semantics, data placement policies, and multi-region topology requires dedicated learning beyond standard RDBMS knowledge.
CAP Theorem
NewSQL databases choose consistency and partition tolerance, sacrificing some availability during network partitions. Spanner uses TrueTime for globally synchronized timestamps, and CockroachDB uses Raft consensus with serializable isolation.
Top Databases
Distributed SQL database inspired by Google Spanner with PostgreSQL wire compatibility, automatic sharding, and multi-region survivability. Built for cloud-native global applications.
Globally distributed relational database with external consistency using TrueTime. The first system to offer both horizontal scalability and strong ACID guarantees at global scale.
Open-source distributed SQL database with MySQL compatibility, supporting HTAP (hybrid transactional/analytical processing) workloads with a columnar storage engine (TiFlash).
Open-source distributed SQL database compatible with PostgreSQL and Cassandra APIs. Offers tunable reads (strong or timeline consistency) and multi-region deployment with geo-partitioning.