turn event streams into (micro) services

The next generation of real-time data processing technology is here. Rapidly transform your event streams into queryable microservices.

Build. Experiment. Launch. Repeat. Pyroclast changes the real-time data game.

What is it?

Pyroclast is a Platform as a Service (PaaS) for transforming streams of event data into queryable microservices. One approach to achieving microservices-on-streams is to store all event data in a database, and then periodically query it from services. This technique becomes inflexible when adapting event data to conflicting business requirements. Querying a centralized database introduces performance issues, which are generally solved via indexes and query caching. While these techniques may solve immediate concerns, they prevent you from leveraging your data for other uses. These problems are exacerbated as data increases in volume, velocity, and variability.

A far more powerful technique inverts this process by materializing views against real-time data streams, instead of directly querying centralized databases. This inversion of thinking enables teams to build various applications that do not depend on each other, eliminating contention across use cases and improving scalability. A streaming centric approach makes time travel easy, allowing applications to replay data streams for speculative queries or correcting past mistakes.

Pyroclast is a fully hosted service for ingesting event data, creating materialized views, and hosting queryable microservices accessible to the rest of your software architecture.

1. Connect event sources

2. Design materialized views

3. Deploy microservices

Capture and emit data from any event source. Send event data directly into our hosted storage, or connect to common storage engines, like Kafka, DynamoDB, and SQL.

A materialized view is a queryable set of results continuously derived from a data stream. Materialized views are a powerful design tool for integrating data streams into your larger software architecture. Build materialized views by combining an intuitive set of functions, filters, and aggregations - or write your own code.

Deploy your materialized view as a fault tolerant microservice. Query your microservice on demand, or have derived results pushed to you in real time. Materialized views support deep parameterization, making it ideal for use cases requiring customization or templating.

Pyroclast Features

Low Latency

Build real-time services with Pyroclast's impressive latency characteristics. Reduce the time between event occurrence and end use consumption to milliseconds.

Strong Consistency

Built for applications that need correctness and consistency, Pyroclast doesn't employ estimation or close-enough guesses about your data. We designed Pyroclast to get the right answer every time.

Replayability

Pyroclast's approach makes it trivial to automatically replay data to correct past mistakes, or develop new services against historical events.

Auditability

The combination of historical data and versioned services makes auditability a breeze. Check assertions about the past and be notified in the future.

Scalability

Pyroclast is built on rigorously tested distributed systems technology, capable of handling billions of events per day. When you're ready to scale up, drag a slider to handle higher throughput.

Deep Introspection

Pyroclast automatically archives problematic data, allowing you to inspect issues and fix your service on the fly.