Ilkka Seppälä
4108f86177
* Changed database implementation. Removed static objects. * Fix Logs * Fix 40 errors from checkstyle plugin run. 139 left)) * Fix CacheStore errors from checkstyle plugin 107 left * Fix last errors in checkstyle. * Fix sonar issues * Fix issues in VALIDATE phase * Fix Bug with mongo connection. Used "Try with resources" * Add test * Added docker-compose for mongo db. MongoDb db work fixed. * Provided missing tests * Comments to start Application with mongo. * Fix some broken links * Remove extra space * Update filename * Fix some links in localization folders * Fix link * Update frontmatters * Work on patterns index page * Work on index page * Fixes according PR comments. Mainly Readme edits. * fix frontmatter * add missing png * Update pattern index.md * Add index.md for Chinese translation * update image paths * update circuit breaker image paths * Update image paths for localizations * add generated puml * Add missing image * Update img file extensions * Update the rest of the EN and ZH patterns to conform with the new website Co-authored-by: Victor Zalevskii <zvictormail@gmail.com>
1.4 KiB
title, category, language, tags
| title | category | language | tags | |
|---|---|---|---|---|
| Master-Worker | Concurrency | en |
|
Also known as
Master-slave or Map-reduce
Intent
Used for centralised parallel processing.
Class diagram
Applicability
This pattern can be used when data can be divided into multiple parts, all of which need to go through the same computation to give a result, which need to be aggregated to get the final result.
Explanation
In this pattern, parallel processing is performed using a system consisting of a master and some number of workers, where a master divides the work among the workers, gets the result back from them and assimilates all the results to give final result. The only communication is between the master and the worker - none of the workers communicate among one another and the user only communicates with the master to get the required job done. The master has to maintain a record of how the divided data has been distributed, how many workers have finished their work and returned a result, and the results themselves to be able to aggregate the data correctly.