Page MenuHomePhabricator

New service request: WDQS Flink based Streaming Updater
Closed, ResolvedPublic

Description

Name: WDQS Streaming Updater
Description: The WDQS Streaming Updater allows generating a stream of updates for WDQS. This service is a replacement for part of the current updater that is deployed on each WDQS server. It allows consolidating the common part of that update process instead duplicating that processing on each node and improves throughput by 5-10x.
Timeline: Q4 2021, hopefully May. Discussions have been ongoing for a few month, so while this timeline is short, it is not unexpected.
Point person: @RKemper as SRE, @dcausse and @Zbyszko as SWE, Search Platform as a team.
Technologies:

  • Flink: a stateful stream processing framework (Java based)
  • Java / JVM
  • Kafka (both as a consumer and a producer)
  • Running on k8s, with some direct dependencies (ability to publish config maps)

Request flow diagram:

streaming-updater-deployment.png (818×1 px, 158 KB)
(Source of the diagram is in the project repo)
Misc:

Details

Event Timeline

Restricted Application added a subscriber: Aklapper. · View Herald Transcript
MPhamWMF renamed this task from [Draft] New service request: WDQ Flink based Streaming Updater to [Draft] New service request: WDQS Flink based Streaming Updater.Apr 19 2021, 7:08 PM

Change 681585 had a related patch set uploaded (by Gehel; author: Gehel):

[wikidata/query/rdf@master] Deployment diagram of WDQS Streaming Updater.

https://gerrit.wikimedia.org/r/681585

akosiaris triaged this task as Medium priority.Apr 21 2021, 11:19 AM
akosiaris moved this task from Inbox to In progress on the Service-deployment-requests board.
Gehel renamed this task from [Draft] New service request: WDQS Flink based Streaming Updater to New service request: WDQS Flink based Streaming Updater.Apr 21 2021, 11:55 AM
Gehel updated the task description. (Show Details)

Change 681585 merged by jenkins-bot:

[wikidata/query/rdf@master] Deployment diagram of WDQS Streaming Updater.

https://gerrit.wikimedia.org/r/681585

Gehel claimed this task.

The WDQS Flink based Streaming Updater is now in production, let's close this ticket.