Package and access data on Amazon S3 in a squash manner
-
Updated
Aug 6, 2023 - Rust
Package and access data on Amazon S3 in a squash manner
A file uploader specialized for uploading many small files onto HDFS
PySpark script to aggregate small parquet files in a prefix into larger files. Designed to be run on AWS Glue
Production-grade Zstd dictionary compression microservice for small JSON APIs. Solves frame overhead & schema drift with trained dictionaries, versioned registries, and dual-signal drift monitoring. 75%+ savings on disjoint holdouts.
Read-only "small-file tax" analyzer for Delta Lake. Reads only the _delta_log to reconstruct the per-partition file-size distribution, then prices the wasted compute and S3 requests in dollars using real Spark internals (openCostInBytes, maxPartitionBytes) and emits a compaction plan with ROI and payback plus the OPTIMIZE SQL. Never runs OPTIMIZE.
Command Line Hangman/Word Guessing game.
Read-only Delta Lake table health scanner with safe fix plans for Databricks, Python, and PySpark.
File merge action to merge files in HDFS or local filesystem
antfs: a distributed blob storage system in Rust, purpose-built for massive small files. Inspired by Ambry / SeaweedFS / CubeFS — append-only log + index for O(1) small-file I/O, built-in Raft control plane, REST API, admin dashboard.
Fast multithreaded archiver for massive numbers of small files, optimized for cloud object storage
To associate your repository with the small-files topic, visit your repo's landing page and select "manage topics."