Edge HTTP to S3

The product is a managed ingest service that accepts JSON over HTTP and MQTT and writes compressed segments and Parquet files directly to a customer's S3 bucket, removing the need to operate brokers, clusters, or custom ETL code to land streaming data for querying. It is sold to developers and data teams in business-to-business environments that need to collect event or device data into cloud storage. It is positioned as an alternative to Kafka and Kinesis that delivers data directly to S3 without brokers or glue code.

Key features

• HTTP and MQTT to S3 ingest
• Dedicated edge endpoints on isolated VMs
• 15+ global edge regions
• Autoscaling by RPS and latency
• Multi-format output in one commit
• Compressed WAL segments
• Raw Parquet with JSON payloads
• Typed Parquet via view definitions
• Declarative YAML view definitions
• JSONPath extraction from nested payloads
• Hive-style date partitioning
• Row filtering at materialisation
• Schema Autopilot AI assistance
• Dead letter queues per view
• Error rate metrics
• Schema preview
• Durable NVMe WAL with fsync
• Sub-10ms p95 acknowledgment
• Automatic crash recovery
• Ordered gapless S3 delivery