drill-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Vitalii Diravka <vita...@apache.org>
Subject Re: Slow query execution in Drill Embedded
Date Wed, 20 Jan 2021 14:54:08 GMT
Hello Jonathan,

Did you try to execute both logical and physical plans? Does the query
involve any aggregation, join or sort operators?
Could you provide the query here, please? Did you try to exclude some
fields from the query (to identify what fields cause the delay).

Kind regards
Vitalii


On Sat, Jan 2, 2021 at 8:55 AM Jonathan Shraga
<jonathan_shraga@yahoo.com.invalid> wrote:

> Hello,
> Release: 1.18
> Evaluating Drill for a large Business Reporting project.Using
> drill-embedded to query Parquet data hosted in s3.Query execution on small
> objects/tables (<200K) takes about 9 sec.Looking at query execution
> statistics, about 7-8 seconds are spent on query planning alone.Enabled the
> metastore (Iceberg), ANALYZED and REFRESH METADATA for the table.Browing
> the metadata store directory shows the table's metadata files (parquet
> format), andquering INFORMATION_SCHEMA works as expected.
> However, query execution remains slow and running EXPLAIN still takes 7-8
> seconds.
> Any tips will be greatly appreciated.
> - Jonathan

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message