Skip to content

SYSTEM Cited by 1 source

Apache Yunikorn

Apache Yunikorn is an open-source resource scheduler originally developed by Cloudera and donated to the Apache Software Foundation — an alternative to Kubernetes' default scheduler for batch + data + ML workloads. Yunikorn provides hierarchical queues, resource fairness, gang scheduling, and job-level (not just pod-level) scheduling semantics; it plugs into Kubernetes as a scheduler replacement.

Upstream: https://yunikorn.apache.org/.

Role in this wiki

Introduced as the scheduling framework underneath Pinterest's Moka Big Data Processing platform. In the Piqama integration, each Pinterest project maps to a Yunikorn queue; quota values from Piqama are translated into Yunikorn queue configuration (memory + vcore floor / ceiling + max concurrent apps) by a Yunikorn Config Updater.

Seen in

  • sources/2026-02-24-pinterest-piqama-pinterest-quota-management-ecosystem — named as the resource-scheduling framework backing Moka; integrated with Piqama's capacity-quota control plane via the Yunikorn Config Updater polling Piqama for quota values and rewriting queue configs. Canonical wiki instance of Yunikorn as an enforcement substrate for a separately-operated quota-management platform.
Last updated · 319 distilled / 1,201 read