Silent data loss occurs in 100% of SIGKILL cases between upload and commit in Spark-on-AWS-Lambda with open table formats, and SafeWriter prevents it by triggering format-native rollbacks via a pre-timeout watchdog.
Gray,Notes on Data Base Operating Systems
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.DC 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Characterizing and Fixing Silent Data Loss in Spark-on-AWS-Lambda with Open Table Formats
Silent data loss occurs in 100% of SIGKILL cases between upload and commit in Spark-on-AWS-Lambda with open table formats, and SafeWriter prevents it by triggering format-native rollbacks via a pre-timeout watchdog.