Tinybird supports multiple ingestion methods: the Events API for JSON/NDJSON over HTTP, a fully managed Kafka connector for streaming data, S3 and GCS connectors for batch imports, local or remote file uploads via the CLI, and table functions to read from PostgreSQL, MySQL, and Apache Iceberg. There are also step-by-step guides for 25+ sources including Snowflake, BigQuery, MongoDB, Stripe, Auth0, OpenTelemetry, Vercel, GitHub, and more.

Data ingested via the Events API is available for queries in under 4 seconds. Kafka connector latency depends on your topic throughput but typically achieves sub-second flushing. S3 and GCS connectors support automatic or on-demand ingestion modes.

All ingestion is fully managed. You don't deploy, scale, or maintain anything. Tinybird handles consumer autoscaling for Kafka, buffering and compaction for the Events API, and polling for object storage connectors. You provide credentials and configuration; Tinybird handles everything else.

Tinybird provides multiple reliability mechanisms. The Events API supports idempotent ingestion with a 5-hour retry window using data hashes to prevent duplicates. The Kafka connector uses a quarantine system (dead letter queue) for messages that fail schema validation, preserving the original message plus error details. All connectors include built-in retries and ingestion protection mechanisms.

The Events API accepts JSON and NDJSON with optional Gzip compression. The Kafka connector supports JSON and Avro (with Schema Registry). S3 and GCS connectors support CSV, NDJSON, and Parquet with multiple compression formats including Gzip, Snappy, LZ4, Zstd, and Brotli. Local and remote file uploads support CSV, NDJSON, and Parquet.

Use the FORWARD_QUERY instruction in your .datasource file to define how existing data transforms to your new schema. Tinybird applies schema changes with zero downtime during deployment. The Kafka connector supports Schema Registry for decoding evolving Avro schemas.

Yes. Tinybird Sinks let you export processed data to S3, GCS, and Kafka. This enables flexible data distribution: you can build complete data circulation loops where data flows in from one source, gets transformed with SQL, and flows out to another destination.

Tinybird offers plan-based pricing with Free, Developer, SaaS, and Enterprise tiers. Each plan includes QPS, vCPU minutes, and storage allocations. All ingestion methods — Events API, Kafka, S3, GCS, and file uploads — are included in every plan with no per-connector fees. The Developer plan starts at $25/month. Visit our pricing for full details.

Yes. The Tinybird TypeScript SDK and Python SDK let you define data sources, connections, pipes, and endpoints as code with full type inference. You can define your entire ingestion pipeline programmatically and deploy it with the CLI.

No. Tinybird supports environment branching with isolated configuration for development, staging, and production. Secrets, connection credentials, and consumer group IDs are scoped per environment. Use tb dev for local development and deploy to production with CI/CD.

Tinybird's Kafka connector is fully managed with automatic consumer autoscaling, offset management, failure recovery, and high availability. Building your own requires deploying consumers, handling backpressure, managing offsets, and monitoring everything. For a detailed comparison, see the dedicated Kafka Connector page.