[MCOL-4888] improve cpimport operating on wide tables Created: 2021-10-04  Updated: 2023-12-19  Resolved: 2023-12-19

Status: Closed
Project: MariaDB ColumnStore
Component/s: PrimProc
Affects Version/s: None
Fix Version/s: 23.10

Type: New Feature Priority: Major
Reporter: Gregory Dorman (Inactive) Assignee: Roman
Resolution: Won't Fix Votes: 0
Labels: None

Sprint: 2021-11, 2021-12, 2021-13, 2021-14, 2021-15

 Description   

drrtuy speaking:

make a cpimport optimization that will benefit bulk insertion for tables with many columns. In short cpimport saves some extra data to be able to rollback. It is about 200kB per column file so for 200 column table this is 40MB that is not so big for the modern disks but slows down the operation. I recall a prospect who tolld that ClickHouse ingests data faster then MCS and this metadata was the reason why MCS was slow ingesting.



 Comments   
Comment by Roman [ 2023-12-19 ]

CH ingests faster from CSV and it does this b/c it leverages parallel CSV read.

Generated at Thu Feb 08 02:53:46 UTC 2024 using Jira 8.20.16#820016-sha1:9d11dbea5f4be3d4cc21f03a88dd11d8c8687422.