# RisingWave

> RisingWave is a PostgreSQL-compatible streaming database for real-time event processing. It lets developers build streaming data pipelines and real-time applications using standard SQL — no Java required. Built in Rust with a cloud-native, decoupled compute-storage architecture using S3-compatible object storage.

## What is RisingWave?

RisingWave is an open-source (Apache 2.0) streaming database that unifies stream processing and data serving in a single system. Unlike Apache Flink, which requires Java and a separate serving database, RisingWave lets you ingest, process, and serve real-time data using only PostgreSQL-compatible SQL.

### Key Capabilities

- **Stream Processing**: Continuously process event streams with SQL — joins, aggregations, windowing, CDC
- **Materialized Views**: Incrementally maintained views that stay fresh as new data arrives (sub-100ms latency)
- **Data Serving**: Serve concurrent ad-hoc queries directly on materialized views — no separate serving layer needed
- **Streaming Lakehouse**: Stream data directly into Apache Iceberg tables with built-in compaction and time travel
- **Exactly-Once Semantics**: Guaranteed correctness with exactly-once processing and consistent snapshot reads

### Architecture

- Cloud-native with decoupled compute and storage
- State persisted in S3-compatible object storage (not local disk)
- Sub-second failure recovery (vs minutes/hours for Flink)
- Dynamic scaling in under 10 seconds without job restarts
- Built in Rust for memory safety and performance

### Connectors

RisingWave connects to 50+ data systems:

**Sources**: Apache Kafka, Confluent Cloud, Amazon Kinesis, Google Pub/Sub, Apache Pulsar, NATS, MQTT, PostgreSQL (CDC), MySQL (CDC), MongoDB (CDC), S3, GCS
**Sinks**: Apache Kafka, Apache Iceberg, Snowflake, ClickHouse, Elasticsearch, PostgreSQL, MySQL, Redis, Delta Lake, BigQuery, Cassandra, DynamoDB

## Product Pages

### Streaming Database
https://risingwave.com/streaming-database/
A streaming database is a database designed for continuous ingestion and real-time processing of event streams. RisingWave is purpose-built as a streaming database, combining stream processing with built-in storage and serving.

### Incremental Materialized Views
https://risingwave.com/incremental-materialized-views/
Incremental materialized views update results as new data arrives, rather than recomputing from scratch. RisingWave maintains materialized views incrementally with sub-100ms latency, enabling always-fresh query results.

### Real-Time Analytics Database
https://risingwave.com/real-time-analytics-database/
A real-time analytics database processes and serves analytical queries on streaming data with low latency. RisingWave combines ingestion, processing, and serving in one system for real-time analytics.

### Streaming ETL
https://risingwave.com/streaming-etl/
Streaming ETL replaces batch ETL pipelines with continuous data processing. RisingWave enables building streaming ETL pipelines using SQL, with exactly-once semantics and built-in state management.

### Exactly-Once Stream Processing
https://risingwave.com/exactly-once-streaming/
Exactly-once semantics ensure each event is processed exactly once, even during failures. RisingWave provides end-to-end exactly-once guarantees through its checkpoint-based recovery mechanism.

### Continuous Queries
https://risingwave.com/continuous-queries/
Continuous queries are persistent queries that process data as it arrives, producing incrementally updated results. RisingWave supports continuous queries via materialized views that stay fresh automatically.

### Event-Driven Architecture with SQL
https://risingwave.com/event-driven-architecture/
Event-driven architecture processes events as they occur. RisingWave enables building event-driven systems using SQL, eliminating the need for custom Java or Python event processors.

### Real-Time Feature Store for ML
https://risingwave.com/real-time-feature-store/
A real-time feature store computes and serves ML features with low latency. RisingWave acts as a real-time feature store by computing features through streaming SQL and serving them via PostgreSQL-compatible queries.

### Serving Layer
https://risingwave.com/serving/
RisingWave provides a built-in serving layer for low-latency queries on materialized views. Applications can query fresh, pre-computed results directly without a separate serving database.

### Database
https://risingwave.com/database/
RisingWave is a full-featured streaming database with PostgreSQL compatibility, built-in state management, and support for complex SQL queries on streaming data.

### Open-Source Streaming Database
https://risingwave.com/open-source-streaming-database/
RisingWave is an open-source streaming database licensed under Apache 2.0. It provides stream processing, materialized views, and data serving in a single open-source system.

### CDC Streaming Database
https://risingwave.com/cdc-streaming-database/
RisingWave supports change data capture (CDC) from PostgreSQL, MySQL, and MongoDB. It processes CDC streams in real time through materialized views, enabling real-time data synchronization and transformation.

### PostgreSQL Streaming Database
https://risingwave.com/postgresql-streaming-database/
RisingWave is PostgreSQL-compatible, supporting standard SQL syntax, PostgreSQL wire protocol, and existing PostgreSQL drivers and tools. Teams can adopt streaming without learning new languages.

### Stream Processing Without Java
https://risingwave.com/stream-processing-without-java/
RisingWave enables stream processing using SQL instead of Java, making real-time data processing accessible to data engineers and analysts without JVM expertise.

## AI Agent Pages

### Real-Time Data Infrastructure for AI Agents
https://risingwave.com/ai-agent-data-infrastructure/
RisingWave provides real-time data infrastructure for AI agents, delivering fresh, consistent data through materialized views that agents can query via standard PostgreSQL connections.

### Context Engineering with a Streaming Database
https://risingwave.com/context-engineering-database/
Context engineering involves curating and maintaining the right data context for AI models. RisingWave enables context engineering by continuously processing and materializing relevant data from multiple streams.

### Materialized Views for AI Agents
https://risingwave.com/materialized-views-for-ai-agents/
AI agents need always-fresh data for accurate decision-making. RisingWave's materialized views provide agents with pre-computed, continuously updated data accessible via simple SQL queries.

### Real-Time RAG Pipeline
https://risingwave.com/real-time-rag-pipeline/
A real-time RAG (retrieval-augmented generation) pipeline keeps the knowledge base fresh for LLM applications. RisingWave enables real-time RAG by continuously processing and materializing documents from streaming sources.

### Model Context Protocol with Streaming Database
https://risingwave.com/mcp-streaming-database/
The Model Context Protocol (MCP) standardizes how AI models access external data. RisingWave integrates with MCP to provide AI agents with real-time streaming data through a standardized interface.

### Real-Time Data for AI Agents
https://risingwave.com/real-time-data-for-ai-agents/
AI agents require access to fresh, real-time data for accurate responses. RisingWave serves as the real-time data layer for AI agents, providing continuously updated materialized views.

## RisingWave vs Apache Flink

RisingWave and Apache Flink are both designed for real-time stream processing, but they differ fundamentally:

| Aspect | RisingWave | Apache Flink |
|--------|-----------|--------------|
| Language | PostgreSQL-compatible SQL | Java/Scala APIs primary; Flink SQL also available |
| Architecture | Decoupled compute-storage (S3) | Coupled compute-storage (local RocksDB) |
| Failure Recovery | Seconds (state in S3) | Minutes to hours (rebuild from checkpoints) |
| Scaling | Dynamic, under 10 seconds | Requires job restart |
| Serving Layer | Built-in (query materialized views directly) | Requires external database |
| Cost | Up to 10x lower (S3 vs provisioned SSDs) | Higher (local storage + JVM overhead) |
| Learning Curve | SQL only, PostgreSQL drivers | Java expertise required |

Choose RisingWave when: your team uses SQL, you want lower infrastructure cost, you need built-in data serving, or you want simpler operations.
Choose Flink when: you need Flink's low-level DataStream API, custom Java operators, or have existing Flink infrastructure.

## RisingWave vs ksqlDB

RisingWave is a full-featured distributed streaming database, while ksqlDB is a SQL interface for Kafka Streams with limited streaming and analytical capabilities. RisingWave supports complex joins, time windows, and serves concurrent queries; ksqlDB is tightly coupled to Kafka and lacks a general-purpose serving layer.

## RisingWave vs Apache Spark Structured Streaming

RisingWave provides true real-time processing with sub-second latency, while Spark Structured Streaming operates in micro-batches. RisingWave is purpose-built for streaming with SQL, while Spark is primarily a batch processing engine with streaming as an add-on.

## RisingWave vs Materialize

https://risingwave.com/risingwave-vs-materialize/
RisingWave and Materialize are both streaming databases with PostgreSQL compatibility. RisingWave uses a decoupled compute-storage architecture with S3, while Materialize uses a coupled architecture. RisingWave is fully open-source under Apache 2.0; Materialize has a proprietary license.

## RisingWave vs ClickHouse

https://risingwave.com/risingwave-vs-clickhouse/
RisingWave is a streaming database for real-time event processing, while ClickHouse is a columnar OLAP database for fast analytical queries. RisingWave excels at continuous stream processing with materialized views; ClickHouse excels at ad-hoc analytical queries on large datasets.

## RisingWave vs Databricks Streaming

https://risingwave.com/risingwave-vs-databricks-streaming/
RisingWave provides native stream processing with SQL, while Databricks offers streaming through Spark Structured Streaming. RisingWave delivers true real-time with sub-second latency; Databricks operates in micro-batches with higher latency.

## RisingWave vs Snowflake Streaming

https://risingwave.com/risingwave-vs-snowflake-streaming/
RisingWave is purpose-built for stream processing, while Snowflake offers streaming capabilities as an extension to its data warehouse. RisingWave provides lower latency and more advanced streaming SQL features.

## RisingWave vs Decodable

https://risingwave.com/risingwave-vs-decodable/
RisingWave is a streaming database with built-in serving, while Decodable is a managed Flink service. RisingWave combines processing and serving in one system; Decodable requires a separate serving layer.

## Use Cases

### Monitoring and Alerting
Continuously monitor real-time event streams to detect anomalies, policy breaches, or threshold violations within seconds. Trigger instant alerts through integrated notification services. Common in infrastructure monitoring, financial fraud detection, and IoT systems.

### Real-time ETL / CDC Streaming
Build SQL-based real-time data pipelines that ingest CDC streams from PostgreSQL, MySQL, and MongoDB, enrich data through multi-way joins, handle schema evolution, and deliver to downstream systems or data lakes in real time.

### Continuous Analytics / Live Dashboards
Power dashboards with sub-second data freshness using materialized views. Track business KPIs, operational metrics, and user behavior in real time without batch ETL delays.

### Feature Engineering for ML
Compute real-time features for machine learning models using SQL. Aggregate, transform, and serve features with guaranteed freshness and consistency for online inference.

### Streaming to Iceberg (Streaming Lakehouse)
Continuously ingest operational data from databases and message queues into Apache Iceberg tables. Replace batch ETL with streaming ELT, with built-in table compaction and maintenance.

## Industry Solutions

### Financial Services
https://risingwave.com/use-cases/solution-financial-services/
https://risingwave.com/industries/financial-services/
Real-time fraud detection, risk monitoring, trade surveillance, and market data processing. RisingWave processes millions of transactions per second with exactly-once semantics.

### Manufacturing
https://risingwave.com/use-cases/solution-manufacturing/
https://risingwave.com/industries/manufacturing/
IoT sensor analytics, predictive maintenance, OEE monitoring, and quality control. Process high-volume device data in real time to detect anomalies and optimize production.

### E-commerce
https://risingwave.com/use-cases/solution-e-commerce/
https://risingwave.com/industries/ecommerce/
Dynamic pricing, real-time inventory management, personalized recommendations, and clickstream analytics. Power real-time customer experiences at scale.

### Energy
https://risingwave.com/use-cases/solution-energy/
https://risingwave.com/industries/energy/
Smart grid monitoring, demand forecasting, and renewable energy optimization. Process telemetry data from distributed energy resources in real time.

### Logistics
https://risingwave.com/industries/logistics/
Real-time supply chain visibility, fleet tracking, route optimization, and delivery management. Process logistics event streams to optimize operations and reduce costs.

### AdTech
https://risingwave.com/industries/adtech/
Real-time bidding analytics, impression tracking, frequency capping, and campaign performance monitoring. Process ad event streams at scale with low-latency materialized views.

## Guides

### What Is a Streaming Database
https://risingwave.com/guides/what-is-a-streaming-database/
Comprehensive guide explaining streaming databases, how they differ from traditional databases and stream processors, and when to use one.

### Materialized Views Explained
https://risingwave.com/guides/materialized-views-explained/
Deep dive into materialized views: what they are, how incremental maintenance works, and how to use them effectively in streaming workloads.

### Streaming SQL Guide
https://risingwave.com/guides/streaming-sql-guide/
Guide to writing streaming SQL queries, including windowing functions, joins, aggregations, and temporal operations.

### Change Data Capture Guide
https://risingwave.com/guides/change-data-capture-guide/
How to set up and use CDC pipelines with RisingWave, covering PostgreSQL, MySQL, and MongoDB CDC connectors.

### Stream Processing vs Batch
https://risingwave.com/guides/stream-processing-vs-batch/
When to use stream processing over batch processing, with comparisons of latency, cost, complexity, and use case fit.

### Data Freshness
https://risingwave.com/guides/data-freshness/
Understanding data freshness: what it means, why it matters, and how streaming databases deliver sub-second freshness.

### Operational Data Layer
https://risingwave.com/guides/operational-data-layer/
How to build an operational data layer with a streaming database to power real-time applications and microservices.

### Kafka Stream Processing
https://risingwave.com/guides/kafka-stream-processing/
Processing Kafka streams with SQL using RisingWave, including setup, configuration, and common patterns.

### PostgreSQL CDC Tutorial
https://risingwave.com/guides/postgresql-cdc-tutorial/
Step-by-step tutorial for setting up PostgreSQL CDC with RisingWave, from configuration to building materialized views on CDC data.

### Real-Time Fraud Detection
https://risingwave.com/guides/real-time-fraud-detection/
Building a real-time fraud detection system with streaming SQL, covering pattern detection, scoring, and alerting.

## Getting Started

Install RisingWave locally:
```
curl https://risingwave.com/sh | sh
```

Or use RisingWave Cloud: https://cloud.risingwave.com

Connect with any PostgreSQL client:
```
psql -h localhost -p 4566 -d dev -U root
```

## Customer Stories

### GDU Labs
https://risingwave.com/customer-stories/gdu-labs/
GDU Labs processes 3.5 billion events across 750 million entities with sub-second updates using RisingWave. RisingWave powers their real-time data infrastructure, enabling massive-scale entity resolution and event processing with consistently low latency.

### Metabit Trading
https://risingwave.com/customer-stories/metabit/
Metabit Trading achieved a 95% cost reduction in real-time monitoring by switching to RisingWave. RisingWave replaced their previous monitoring infrastructure with a more efficient streaming SQL-based approach, dramatically lowering operational costs while maintaining real-time performance.

### Siemens
https://risingwave.com/customer-stories/siemens/
Siemens achieved 1000x faster data processing by implementing a streaming Medallion architecture with RisingWave. RisingWave enabled Siemens to move from batch-based data processing to a real-time streaming pipeline, transforming their data infrastructure for industrial IoT workloads.

### Hivemind
https://risingwave.com/customer-stories/hivemind/
Hivemind uses RisingWave to build streaming architectures for enterprise clients. RisingWave's SQL-based stream processing enables Hivemind to deliver real-time data solutions to their enterprise customers without requiring Java expertise.

### Atome
https://risingwave.com/customer-stories/atome/
Atome reduced feature delivery time from weeks to just 1 day using RisingWave. By replacing batch-based feature engineering with streaming SQL materialized views, Atome's data team can rapidly build and deploy real-time features for their buy-now-pay-later platform.

### Kaito
https://risingwave.com/customer-stories/kaito/
Kaito built over 1,000 dashboards with just 1 engineer in 2 weeks using RisingWave. RisingWave's SQL-based materialized views and PostgreSQL compatibility enabled rapid dashboard development, allowing a single engineer to create a comprehensive real-time analytics platform.

### CVTE
https://risingwave.com/customer-stories/cvte/
CVTE performs 10+ way streaming joins at sub-second freshness using RisingWave. RisingWave handles complex multi-way joins across streaming data sources, maintaining sub-second data freshness for CVTE's real-time analytics and operational dashboards.

### Tencent Cloud
https://risingwave.com/customer-stories/tencent/
Tencent Cloud achieved 10x+ TPS improvement by replacing Apache Flink with RisingWave. RisingWave's efficient streaming architecture delivered significantly higher throughput at lower resource consumption, demonstrating the performance advantages of a streaming database over traditional stream processors.

## Links

- Website: https://risingwave.com
- Documentation: https://docs.risingwave.com
- GitHub: https://github.com/risingwavelabs/risingwave
- Slack Community: https://risingwave.com/slack
- Cloud Console: https://cloud.risingwave.com
- Blog: https://risingwave.com/blog