google cloud dataflow - Are template tables supported in BigQuery for bulk import? -
there several options loading data bigquery: e.g. bulk import gcs, streaming , others.
in many cases, 1 needs shard data being loaded, e.g. date, or arbitrary key, in order produce smaller tables faster query, or around per-table import quotas.
recently, new feature introduced, template tables, makes such sharding easy streaming: specify suffix of table name want stream to, on per-record basis.
is bigquery feature available other import modes, importantly import gcs? useful importing large amounts of data bigquery in sharded way, common use case e.g. when using cloud dataflow batch jobs.
no, template tables not available bulk import @ time; rationale since bulk import can create tables side-effect, wouldn't necessary.
for streaming imports, semantics bit trickier. since streaming insert requests don't specify schema, if destination table doesn't exist, bigquery doesn't know desired schema of table should be. template tables allow streaming system desired schema somewhere else.
for bulk loads, however, schema included part of request, or can inferred data, template tables don't make sense.
all said, we're aware management of multiple sharded tables inconvenient, , hope have improvements ready soon.
Comments
Post a Comment