RunPod discloses an in-house CDC event-streaming platform; billing is migrating off database polling onto it
An engineering post details RunpodEventsStream, an in-house change-data-capture pipeline built so billing and notifications can stop polling the database. Analytics reads it today; the billing consumer is still in development. Snowflake is the one bought piece, on the downstream analytics layer.
RunPod’s engineering blog published “The binlog is the source of truth: How we built event streaming infrastructure at Runpod” (August 25, 2026), disclosing that billing detected state changes by polling the database: “Billing polled the database to detect state changes… Every one of those loops introduced lag.”
The fix is a homegrown CDC pipeline — two Kinesis-backed streams, RunpodEventsStream and CDCEventsStream, fed from MySQL/Vitess binlogs via a CDC service RunPod wrote itself rather than adopting the usual Debezium/Kafka-Connect stack (“so we built the CDC service on VStream directly”). Billing is named as a first consumer, but not a live one: “the billing service wants to know when a billable event occurred, in a stable format, with the required fields, not a stream of raw row diffs it has to interpret,” while “Operational consumers like billing and notifications still run on the older polling paths, and moving them onto events is where this is headed.” The post describes the billing consumer as “currently in development” and its own results table records the billing guarantee as “Migration to events in progress.”
The one piece RunPod did buy is the downstream analytics warehouse: “its managed Snowflake connector turned out to be genuinely easy to stand up … it is also the stream our Snowflake analytics pipeline reads today.” Snowflake reads the same schema-validated event stream that billing is being migrated onto, instead of running its own separate extraction.
This reinforces a pattern already on RunPod’s page: it owns the metering/billing core (its per-second credit-ledger billing) while buying commodity infrastructure — payments via Stripe, analytics via Snowflake — around it.
RunPod added an Enterprise nav entry and a dedicated runpod.io/enterprise page that documents the enterprise contract shape for the first time: reserved baseline capacity at committed-use pricing plus usage-based burst above the baseline, billed post-paid on one consolidated invoice, with contractual SLAs, SOC 2 Type II / HIPAA / GDPR, org-level RBAC and SSO, and clusters scaling from 200+ on demand to 10,000 reserved. The same surface launches Runpod Overdrive, a vLLM serving-optimization layer for Serverless sold on an outcome guarantee ("If we don't beat your baseline, you pay nothing") claiming up to 3.5x faster token streaming, 3.28x faster time to first token and 2.45x higher throughput. No enterprise dollar figures and no Overdrive price are published, and every self-serve rate held at its 2026-07-14 level. The pricing page also gained a visible "Updated July 17, 2026" freshness stamp.