Performance with Stream:Yes

Help
1

What is the meaning of Stream:Yes in below explain. Apart from that can i say that bloom filter is always better than the normal filter.

EXPLAIN SELECT T1.col_char, T3.id from T1,T3 where T1.id = T3.id and T3.id = 5 – Join on non Shard Key Columns and Filter on Shard Key column (Distributed Join)

Gather partitions:all est_rows:1 alias:remote_0

Project [r0.col_char, T3.id] est_rows:1 est_select_cost:2

Filter [T3.id = 5]

NestedLoopJoin

|—IndexRangeScan test.T3, SHARD KEY id (id) scan:[id = r0.id] est_table_rows:2 est_filtered:1

TableScan r0 storage:list stream:yes est_table_rows:1

Repartition [T1.col_char, T1.id] AS r0 shard_key:[id] est_rows:1

Filter [T1.id = 5]

TableScan test.T1 est_table_rows:2 est_filtered:1

2

TableScan r0 storage:list stream:yes est_table_rows:1

represents an internal temp table used for processing this query that is never materialized. The rows from producer(s) are streamed to the consumer(s).

Bloom filters are used typically to speed up hash joins.