Hi everyone,
Last year, we started working on using other types of storage for more reliance during Spark shuffle as tracked by this jira ticket. We’ve compiled a document outlining the progress we’ve made since our last update in December. Specifically, this includes: A summary of the work done so far A list of problems with an initial API and prototype that we’ve been iterating on Links to some initial benchmarking results A detailed proposal for the next phase of the project, including: The new proposed API Detailed project milestones A standard process for submitting, reviewing, and testing PRs We ask that anybody who is familiar or interested in this work to review and comment on the document, or by responding to this email thread or on the jira ticket. Thanks! Yifei
smime.p7s
Description: S/MIME cryptographic signature