[SPARK-19487] Low latency execution for Spark

• Type: Umbrella
• Status: Resolved
• Priority: Major
• Resolution: Incomplete
• Affects Version/s: 2.1.0
• Component/s: ML, Scheduler, Spark Core, Structured Streaming
• Labels: bulk-closed
• Assignee: Unassigned
• Reporter: Shivaram Venkataraman
• Created: 07/Feb/17 06:49
• Updated: 17/May/20 17:48
• Resolved: 21/May/19 04:12

Description

This JIRA tracks the design discussion for supporting low latency execution in Apache Spark. The motivation for this comes from need to support lower latency stream processing and lower latency iterations for sparse ML workloads.

Overview of proposed design (in the format of Spark Improvement Proposal) is at https://docs.google.com/document/d/1m_q83DjQcWQonEz4IsRUHu4QSjcDyqRpl29qE4LJc4s/edit?usp=sharing

Source code prototype is at: https://github.com/amplab/drizzle-spark

Lets use this JIRA to discuss high level design and we can create subtasks as we break this down into smaller PRs.

This is joint work with kayousterhout