Details
Description
Uber JIRA to track things needed for Azure Data Lake to be considered stable and adl:// ready for wide use.
Based on the experience with other object stores, the things which usually surface once a stabilizing FS is picked up and used are
- handling of many GB files, up and down, be it: efficiency of read, when the writes take place, file leakage, time for close() and filesystem shutdown
- resilience to transient failures
- reporting of problems/diagnostics
- security option tuning
- race conditions
- differences between implementation and what actual applications expect
Attachments
Issue Links
- is cloned by
-
HADOOP-14764 Über-jira adl:// Azure Data Lake Phase II: Performance, Resilience and Testing
-
- Resolved
-
There are no Sub-Tasks for this issue.