lakebench-k8s 1.0.6

Deploy and benchmark lakehouse stacks on Kubernetes

Lakebench

A/B testing for lakehouse architectures on Kubernetes.

Deploy a complete lakehouse stack from a single YAML, run a medallion pipeline at any scale, and get a scorecard you can compare across configurations.

Why Lakebench?

• Compare stacks. Swap catalogs (Hive, Polaris), query engines (Trino, Spark Thrift, DuckDB), and table formats -- same data, same queries, different architecture. Side-by-side scorecard comparison.
• Test at scale. Run the same workload at 10 GB, 100 GB, and 1 TB to find where throughput plateaus or resources saturate on your hardware.
• Measure freshness. Continuous mode streams data through the pipeline and benchmarks query performance under sustained ingest load.

Quick Start

pip install lakebench-k8s

Pre-built binaries (no Python required) are available on GitHub Releases.

lakebench init --interactive        # generate config with S3 prompts
lakebench validate lakebench.yaml   # check config + cluster connectivity
lakebench deploy lakebench.yaml     # deploy the stack
lakebench generate lakebench.yaml --wait   # generate test data
lakebench run lakebench.yaml        # run pipeline + benchmark
lakebench report                    # view HTML scorecard
lakebench destroy lakebench.yaml    # tear down everything

The recipe field selects your architecture in one line. The scale field controls data volume.

# lakebench.yaml (minimal)
deployment_name: my-test
recipe: hive-iceberg-spark-trino   # or polaris-iceberg-spark-duckdb, etc.
scale: 10                          # 1 = ~10 GB, 10 = ~100 GB, 100 = ~1 TB
s3:
  endpoint: http://s3.example.com:80
  access_key: ...
  secret_key: ...

Eight recipes are available -- see Recipes for the full list.

What You Get

After lakebench run completes, the terminal prints a scorecard:

 ─ Pipeline Complete ──────────────────────────────
  bronze-verify         142.0 s
  silver-build          891.0 s
  gold-finalize         234.0 s
  benchmark              87.0 s

  Scores
    Time to Value:        1354.0 s
    Throughput:           0.782 GB/s
    Efficiency:           3.41 GB/core-hr
    Scale:                100.0% verified
    QpH:                  2847.3

  Full report: lakebench report
 ──────────────────────────────────────────────────

lakebench report generates an HTML report with per-query latencies, bottleneck analysis, and optional platform metrics (CPU, memory, S3 I/O per pod).

How It Works

                    ┌──────────────────────────────────┐
                    │         lakebench.yaml           │
                    └────────────┬─────────────────────┘
                                 │
                    ┌────────────▼─────────────────────┐
                    │   deploy (Kubernetes namespace,   │
                    │   S3 secrets, PostgreSQL, catalog, │
                    │   query engine, observability)     │
                    └────────────┬─────────────────────┘
                                 │
     Raw Parquet ──► Bronze (validate) ──► Silver (enrich) ──► Gold (aggregate)
         S3              Spark                Spark               Spark
                                                                    │
                                                        ┌───────────▼──────────┐
                                                        │  8-query benchmark   │
                                                        │  (Trino / DuckDB /   │
                                                        │   Spark Thrift)      │
                                                        └──────────────────────┘

Prerequisites

• kubectl and helm on PATH
• Kubernetes 1.26+ (minimum 8 CPU / 32 GB RAM for scale 1)
• S3-compatible object storage (FlashBlade, MinIO, AWS S3, etc.)
• Kubeflow Spark Operator 2.4.0+ (or set spark.operator.install: true)
• Stackable Hive Operator for Hive recipes (not needed for Polaris)

Commands

Command	Description
init	Generate a starter config file
validate	Check config and cluster connectivity
info	Show deployment configuration summary
deploy	Deploy all infrastructure components
generate	Generate synthetic data at the configured scale
run	Execute the medallion pipeline and benchmark
benchmark	Run the 8-query benchmark standalone
query	Execute ad-hoc SQL against the active engine
status	Show deployment status
report	Generate HTML scorecard report
recommend	Recommend cluster sizing for a scale factor
destroy	Tear down all deployed resources

See CLI Reference for flags and options.

Component Versions

Component	Version
Apache Spark	3.5.4
Spark Operator	2.4.0 (Kubeflow)
Apache Iceberg	1.10.1
Hive Metastore	3.1.3 (Stackable 25.7.0)
Apache Polaris	1.3.0-incubating
Trino	479
DuckDB	bundled (Python 3.11)
PostgreSQL	17

All versions are overridable in the YAML config. See Supported Components.

Documentation

• Getting Started -- prerequisites, install, first run
• Configuration -- full YAML reference
• Recipes -- catalog + format + engine combinations
• Running Pipelines -- batch and continuous modes
• Benchmarking -- scorecard and query benchmark
• Architecture -- system design
• Troubleshooting -- common errors and fixes

License

Apache 2.0