cloud-data-cockpit 0.2.2

An IPython widget for browsing, benchmarking, and processing S3 datasets.

Data Cockpit

Data Cockpit is an interactive IPython widget built on top of the Dataplug framework. It enables scientists and engineers to:

• Upload and browse datasets in Amazon S3
• Explore curated public and Metaspace collections
• Benchmark performance to discover optimal batch sizes
• Partition a variety of scientific data types into chunks or batches
• Integrate seamlessly into Jupyter notebooks for elastic, parallel workloads

---

Why Data Cockpit?

Built on Dataplug’s Cloud-Aware Partitioning

Dataplug is a client-side Python framework for dynamic, zero-cost data slicing of unstructured scientific data stored in object stores like S3. It:

• Pre-processes data in a read-only fashion, building lightweight indexes decoupled from the raw objects
• Exploits S3 byte-range reads to parallelize high-bandwidth access across many workers
• Supports a plug-in interface for multiple domains:
  • Generic: CSV, raw text
  • Genomics: FASTA, FASTQ, VCF
  • Geospatial: LiDAR, Cloud-Optimized Point Cloud (COPC), COG
  • Metabolomics: ImzML
• Allows re-partitioning with different strategies without rewriting the original data

What Data Cockpit Adds

While Dataplug focuses on efficient data slicing, Data Cockpit provides an end-to-end Jupyter UI that:

1. Uploads your local files directly into any S3 bucket
2. Browses existing buckets or public datasets from the AWS Open Data Registry
3. Runs benchmarks across a configurable range of batch sizes to find the fastest throughput
4. Processes & partitions your data with one click, displaying progress and results entirely in-notebook
5. Retrieves partitions via get_data_slices(), which returns the DataPlug data slices (metadata) for downstream processing

---

Installation

pip install cloud-data-cockpit

Or install both Data Cockpit and geospatial extras together:

pip install cloud-data-cockpit[geospatial]