A company centralizes VPC flow logs from many accounts into a single S3 bucket. Logs are stored as text files compressed with gzip and must be retained indefinitely. Analysts run ad hoc Athena queries against these logs, but query performance is degrading as log volume grows. You must both improve query performance and reduce storage usage. Which change will provide the LARGEST performance improvement?

Reconfiguring VPC flow log delivery to use Apache Parquet format and hourly partitions offers the largest performance improvement. Parquet is a columnar storage format that significantly reduces the amount of data Athena needs to scan for queries, leading to faster execution and lower costs. Hourly partitioning further optimizes queries by allowing Athena to prune irrelevant data, scanning only the partitions containing the requested time range. Decompressing and recompressing with bzip2 might slightly reduce storage but won't fundamentally improve query performance as the data remains row-oriented. S3 Transfer Acceleration speeds up uploads, not query performance, and S3 Intelligent-Tiering is for cost optimization, not query speed. A new Athena workgroup or engine version 2 might offer minor improvements but won't address the core issue of inefficient data storage and organization as effectively as Parquet and partitioning.