What Are Streams?
A stream is the destination of a log record after being processed by a pipeline. This is the actual collection that you will query/visualize using filters.
Every field can be indexed, allowing fast retrieval of log records matching some filter.
A field's value is embedded in its index key, and storage backends cap the size of a key (FoundationDB rejects keys larger than 10 KB). When an indexed field's value is too large to fit in an index key, FlowG stores the record but skips indexing that value — it does not reject the record.
Such a record is still returned by time-range queries, but it will not match an exact-value filter on that field, and its value will not appear in the field's value suggestions.
With typical (short) stream and field names, values up to roughly 7 KB are always indexed. The exact threshold is deterministic but depends on the backend and on the lengths of the stream and field names — see the storage design for the formula. Avoid indexing free-form, high-cardinality fields such as message bodies.