Skip to main content

​
What is YugabyteDB?

YugabyteDB is a PostgreSQL-compatible, high-performance, cloud-native distributed SQL database. It combines the best of traditional relational databases with the scalability and resilience of NoSQL systems, making it the ideal choice for cloud-native applications that demand both transactional consistency and the ability to handle massive data volumes.
YugabyteDB is 100% open source under the Apache 2.0 license. The source code is available on GitHub.

​
Who should use YugabyteDB?

YugabyteDB is best suited for cloud-native OLTP (real-time, business-critical) applications that need absolute data correctness and require at least one of the following:
  • Horizontal scalability - Your application needs to scale beyond a single database instance
  • High tolerance to failures - You require continuous availability even during infrastructure failures
  • Globally-distributed deployments - Your users are distributed across multiple regions or continents
  • PostgreSQL compatibility - You want distributed SQL without sacrificing PostgreSQL features

​
Why YugabyteDB?

​
PostgreSQL compatibility

YugabyteDB SQL (YSQL) reuses the PostgreSQL query layer, providing compatibility with PostgreSQL 15. You get:
  • Familiar PostgreSQL syntax and semantics
  • Support for PostgreSQL data types, functions, and operators
  • Stored procedures, triggers, and extensions
  • Your existing PostgreSQL tools and drivers work seamlessly
Migrating from PostgreSQL? YugabyteDB offers a seamless lift-and-shift approach. Check out YugabyteDB Voyager for simplified database migration.

​
Built for the cloud

YugabyteDB is designed from the ground up as a cloud-native database:
  • Deploy on any public cloud (AWS, GCP, Azure) or on-premises
  • Native Kubernetes support for containerized environments
  • Multi-cloud and hybrid cloud deployments
  • No dependency on atomic clocks or specialized hardware

​
Resilience and high availability

YugabyteDB is extremely resilient to common outages:
  • Zero data loss (RPO = 0) - No data is lost on failure
  • 3-second recovery (RTO = 3s) - Failed nodes recover automatically
  • Tolerates disk, node, zone, region, and cloud failures
  • Native failover and repair without manual intervention
For a typical multi-zone deployment on a public cloud, YugabyteDB provides an RPO of 0 (no data loss) and an RTO of 3 seconds (recovery time).

​
Distributed by design

Unlike traditional databases with bolted-on replication, YugabyteDB is distributed at its core:
  • Data automatically sharded across nodes using smart partitioning
  • Synchronous replication using Raft consensus protocol
  • Strong consistency for reads and writes
  • ACID transactions across multiple nodes and data centers

​
Key differentiators

Distributed SQL

Full SQL support with distributed ACID transactions, joins, and foreign keys across nodes.

Linear scalability

Scale reads and writes by adding nodes. No application changes required.

Continuous availability

Automatic failover and repair. Survives node, zone, and region failures.

Multi-API support

YSQL (PostgreSQL) and YCQL (Cassandra) APIs on the same database.

​
Architecture overview

YugabyteDB uses a layered architecture that separates concerns: 
┌─────────────────────────────────────────┐
│         Query Layer (YSQL/YCQL)         │  ← PostgreSQL-compatible SQL
├─────────────────────────────────────────┤
│       Distributed Transactions          │  ← ACID across nodes
├─────────────────────────────────────────┤
│    Sharding & Replication (Raft)        │  ← Auto-sharding & consensus
├─────────────────────────────────────────┤
│      Storage Layer (DocDB/RocksDB)      │  ← High-performance storage
└─────────────────────────────────────────┘
1

Query layer

Handles SQL queries using the PostgreSQL query processor. Provides full SQL compatibility.
2

Transaction layer

Implements distributed ACID transactions using hybrid logical clocks for global ordering.
3

Replication layer

Uses Raft consensus for synchronous replication. Ensures strong consistency across replicas.
4

Storage layer

DocDB storage engine built on RocksDB. Optimized for SSDs and NVMe drives.

​
Use cases

YugabyteDB excels in scenarios requiring:

​
Cloud-native applications

Build modern applications that need to scale elastically:
  • Microservices architectures
  • E-commerce platforms
  • SaaS applications
  • Mobile backends

​
Mission-critical systems

Deploy systems that cannot tolerate downtime:
  • Financial services and payments
  • Healthcare systems
  • Telecommunications
  • Gaming platforms

​
Global applications

Serve users worldwide with low latency:
  • Multi-region deployments
  • Geo-distributed data
  • Compliance with data residency requirements
  • Row-level geo-partitioning

​
IoT and time-series

Handle high-volume write workloads:
  • IoT sensor data ingestion
  • Metrics and monitoring
  • Audit logs
  • Real-time analytics

​
Design principles

YugabyteDB was built with several key design goals:

Consistency first

Strong consistency by default. No eventual consistency surprises.

Operational simplicity

Automated deployment, scaling, and self-healing. Minimal ops overhead.

SQL compatibility

Full PostgreSQL compatibility. Use familiar tools and patterns.

Cloud portability

Deploy anywhere. No vendor lock-in.

​
Performance characteristics

YugabyteDB is optimized for:
  • High write throughput - Handle thousands of writes per second per node
  • Low latency reads - Single-digit millisecond latencies with smart client drivers
  • High client concurrency - Support thousands of concurrent connections
  • Large data sets - Store terabytes of data per node
For optimal performance, deploy YugabyteDB on SSDs or NVMe drives. The database is written in C++ and optimized for modern storage hardware.

​
Comparison with other databases

​
vs. PostgreSQL

  • Similarities: Wire-compatible SQL, same syntax and semantics
  • Advantages: Horizontal scalability, automatic failover, multi-region support
  • Trade-off: Slightly higher latency due to distributed consensus

​
vs. Cassandra

  • Similarities: Distributed architecture, high availability
  • Advantages: ACID transactions, SQL support, strong consistency
  • Trade-off: Requires more nodes for fault tolerance (3 vs 1)

​
vs. Cloud-native databases (Aurora, Cloud Spanner)

  • Similarities: Distributed SQL, high availability
  • Advantages: Open source, no vendor lock-in, deploy anywhere
  • Trade-off: Self-managed (though YugabyteDB Aeon provides managed service)

​
Recent innovations

YugabyteDB continues to evolve with cutting-edge features:

​
PostgreSQL 15 compatibility (v2025.1+)

The latest stable release includes:
  • Stored generated columns
  • Foreign keys on partitioned tables
  • Incremental sort optimization
  • Memoization for faster queries

​
AI/ML capabilities

Build intelligent applications with:
  • pgvector extension - Store and query high-dimensional vectors
  • HNSW indexing - Fast similarity search for AI workloads
  • Native support for embedding-based retrieval

​
Enhanced performance

New optimizations in recent releases:
  • Cost-based optimizer (CBO) - Smarter query plans based on statistics
  • Bitmap scan - Efficient multi-index queries
  • Parallel query execution - Leverage multiple CPU cores
  • Auto analyze - Automatic statistics collection
Starting with v2025.2, CBO, auto analyze, and bitmap scan are enabled by default on new clusters for optimal performance.

​
Getting started

Ready to try YugabyteDB? Here’s what to do next:

Quick start

Get a local cluster running in under 5 minutes

Installation

Install YugabyteDB on your preferred platform

Core features

Explore YugabyteDB’s key capabilities

Architecture

Deep dive into the technical architecture

​
Community and support

Join the YugabyteDB community:
Need enterprise support? Check out YugabyteDB Anywhere for self-managed deployments or YugabyteDB Aeon for a fully managed cloud service.

Build docs developers (and LLMs) love

Get started for freeTalk to us