Lightning Table Queries in Confluent Cloud
Lightning Table queries read your topic data with sub-second response times. You can run them in the Confluent Cloud Console, through an HTTP API, or with a Model Context Protocol (MCP) client.
Important
Lightning Tables are rolling out to Confluent Cloud accounts in phases. If you don’t see this feature, it’s not available for your account at this time.
What is a Lightning Table query
A Lightning Table query reads data from the Lightning Table for an Apache Kafka® topic. Confluent Cloud updates the Lightning Table within 100 milliseconds of new data arriving in the topic. The Lightning Table persists independently of the topic’s retention policy. As a result, a Lightning Table query can return data that has already aged out of the topic.
A Lightning Table query runs once and returns a result. In that respect, it works like a snapshot query in Apache Flink®. But a Lightning Table query targets a different performance profile: sub-second response times and high concurrency for point lookups, range scans, and full scans. A snapshot query is better suited to complex, exploratory analysis over historical data.
The following table compares a Lightning Table query with Flink snapshot and streaming queries.
Query type |
How it runs |
Use it for |
|---|---|---|
Lightning Table query |
Runs once and returns a result in less than a second. |
Point lookups, range scans, and full scans from many concurrent clients, such as applications and AI agents. |
Snapshot query |
Runs once, returns results, and exits. |
Complex, exploratory analysis over historical data. |
Streaming query |
Runs continuously and returns results incrementally as new data arrives. |
Processing data continuously as it arrives. |
A Lightning Table query reads the Lightning Table instead of the topic. So its results might differ from the results of a Flink streaming or snapshot query on the same topic.
For more information about Flink queries, see Batch and Stream Processing in Confluent Cloud for Apache Flink.