Kettle optimization is so much.

Copyright notice: This article is an original article for bloggers. It can not be reproduced without permission from bloggers. Https://blog.csdn.net/ClamReason/article/details/49930479

Kettle‘ >/ > Second

2000‘ >/‘ >/

http,get,set

Slower than database

 

 

Scenarios prone to performance problems

 

Query class:

Database query: database query, database connection, insert update

Web

 

Sort class

Sort, merge join (depend on sorting), group (depend on sorting)

 

The key to tuning:Rowset

Rowset“, “run”ktr100” records. If the input is much larger than the output, it means that this step is too late to handle, which is the bottleneck.

 

Rowset10000

Other observation methods: performance diagram is the same as step measurement.

 

How to improve performance

Reasonably increase indexes

Database query: use as much as possible.= >; if it is compared, it will be built.B threads one step. Adjust yourself.

Increase cache: sort class, query class.

Cluster: query class, arithmetic class, sort

Replace other implementations:JavaScript,Java” is.Basic in.kettle inrunning” can not be sent again until the downstream processing is completed. So sorting requires more data, and table input copies records to two downstream streams, one for more data and one for less data. So, deadlock.

Terms of settlement:

 

Leave a Reply

Your email address will not be published. Required fields are marked *