How soumithbabburi/swapo works
SWaPO is a genomic-table physical-design adviser: it observes query workloads and table characteristics, recommends storage/layout changes, and estimates their read-cost impact. Its user-facing path is the swapo advise command; separate benchmark and paper workflows evaluate the method on simulated and real genomic data. The graph includes core decision, data, pruning, execution, and storage-cost layers. Some subsystem connections rely on the SOURCE INDEX’s import/name references or README descriptions; unsampled internals are not treated as verified call paths.
Main entry points: optimizer.py Recommendation engine · model.py Genomic model · cli.py CLI · experiments.py Research experiments
Components 22
User interface
- CLI
swapo/cli.py
Optimization
- Workload analysis
swapo/workload.py - Recommendation engine
swapo/optimizer.py - Read-cost model
swapo/cost.py - Split decision
swapo/threshold.py - Block-size search
swapo/blocksize.py - Row-group sizing
swapo/rowgroups.py - Input provenance
swapo/provenance.py - Freshness costs
swapo/staleness.py
Genomic data
- Genomic model
swapo/model.py - Real-file readers
swapo/real.py - Data simulator
swapo/data.py - Table statistics
swapo/stats.py
Query runtime
- Layout writer
swapo/layout.py - Interval zone map
swapo/zonemap.py - Query executor
swapo/engine.py - Object-store costs
swapo/objstore.py
Measurement
- Research experiments
swapo/experiments.py - Real-data benchmark
swapo/realbench.py - Row-group benchmark
swapo/rgbench.py - Format benchmark
swapo/formatbench.py
Other components
- Analyst
Connections 45
- Analyst → CLI: invokes
README.md - CLI → Recommendation engine: dispatches
swapo/cli.py - CLI → Research experiments: dispatches
swapo/cli.py - CLI → Real-file readers: uses readers
swapo/cli.py - CLI → Layout writer: uses layouts
swapo/cli.py - CLI → Row-group sizing: uses sizing
swapo/cli.py - CLI → Object-store costs: uses costs
swapo/cli.py - Recommendation engine → Workload analysis: analyzes
swapo/optimizer.py - Recommendation engine → Read-cost model: estimates cost
swapo/optimizer.py - Recommendation engine → Row-group sizing: chooses size
swapo/optimizer.py - Recommendation engine → Split decision: evaluates split
swapo/optimizer.py - Recommendation engine → Object-store costs: models storage
swapo/optimizer.py - Recommendation engine → Input provenance: uses lineage
swapo/optimizer.py - Recommendation engine → Freshness costs: checks freshness
swapo/optimizer.py - Recommendation engine → Genomic model: uses table data
swapo/optimizer.py - Recommendation engine → Interval zone map: uses pruning stats
swapo/optimizer.py - Workload analysis → Genomic model: classifies queries
swapo/workload.py - Read-cost model → Genomic model: costs table
swapo/cost.py - Read-cost model → Workload analysis: uses workload
swapo/cost.py - Read-cost model → Interval zone map: models pruning
swapo/cost.py - Split decision → Genomic model: evaluates intervals
- Block-size search → Split decision: uses threshold
swapo/blocksize.py - Row-group sizing → Object-store costs: models requests
swapo/rowgroups.py - Real-file readers → Data simulator: shares data model
swapo/real.py - Real-file readers → Genomic model: builds tables
- Data simulator → Genomic model: builds tables
swapo/data.py - Table statistics → Data simulator: summarizes tables
swapo/stats.py - Table statistics → Genomic model: uses table types
swapo/stats.py - Layout writer → Data simulator: writes table data
swapo/layout.py - Layout writer → Genomic model: builds blocks
swapo/layout.py - Layout writer → Interval zone map: populates map
swapo/layout.py - Query executor → Interval zone map: prunes blocks
swapo/engine.py - Query executor → Genomic model: runs queries
swapo/engine.py - Research experiments → Data simulator: loads data
swapo/experiments.py - Research experiments → Real-file readers: loads real data
swapo/experiments.py - Research experiments → Query executor: evaluates queries
swapo/experiments.py - Research experiments → Layout writer: compares layouts
swapo/experiments.py - Research experiments → Table statistics: records statistics
swapo/experiments.py - Research experiments → Split decision: checks split choice
swapo/experiments.py - Real-data benchmark → Query executor: benchmarks queries
swapo/realbench.py - Real-data benchmark → Real-file readers: loads real data
swapo/realbench.py - Row-group benchmark → Object-store costs: measures requests
swapo/rgbench.py - Row-group benchmark → Row-group sizing: benchmarks sizing
swapo/rgbench.py - Format benchmark → Row-group benchmark: uses benchmarks
swapo/formatbench.py - Format benchmark → Row-group sizing: tests row groups
swapo/formatbench.py