results2text: Benchmark Comparisons

I’ve added a new tool (added in September 2025) to enable you to compare benchmark results that are generated and stored in the database (see here). The tool allows you to choose which results to compare and outputs the results in text or csv. This makes it far easier to analyze multiple runs without juggling local XML files.

What this gives you:

• Direct DB comparisons for Overview, DML totals, per-transaction metrics, and Top Database Wait Events.
• Colourized terminal output.
• CSV export for easy spreadsheet/dashboard analysis.
• A clean, declarative CLI.
• Robust JSON parsing.

Highlights

• Compare multiple runs by ID or pick the most recent N by time.
• List available runs for quick selection.
• See the same metrics you’re used to (TPS, totals, percentiles, waits).
• Output is aligned and colourized for quick scanning.
• Export to CSV files with one flag.

Quick Start

Some examples of the types of commands you might want to use

# Compare last 3 runs (default)
./results2text -cs //host/service -u USER -p PASS

# List latest runs (ID, Time, Name)
./results2text -cs //host/service -u USER -p PASS --list

# Compare specific IDs
./results2text -cs //host/service -u USER -p PASS--ids 122,121,102

# Export CSV (overview.csv, dml.csv, transactions.csv, waits.csv)
./results2text -cs //host/service -u USER -p PASS --last 4 --csv-dir out --csv-prefix tpcc_

Tip: Colours are enabled by default; turn them off with --no-color or when piping output to files.

Command-Line Options

• Connection
  • -cs or --connectstring Oracle DSN, e.g. //host:1521/service
  • -u or--user Database username
  • -p or--password Database password
• Scope and filters
  • -lr or --last N Compare/List last N runs (default: 3)
  • -i or--ids 1,2,3 Compare explicit run IDs (overrides --last)
  • -nl or--name-like TEXT Filter by recording name
  • -l or --list List runs and exit
• Sections and output
  • -tw or --top-waits N Number of top wait events shown (default: 10)
  • -tx or --tx "Name1,Name2" Restrict transaction comparison to these names
  • -nc or --no-color Disable terminal colour
  • -cd or -cs or --csv-dir DIR Write CSV outputs to DIR
  • -cp or --csv-prefix PREFIX Optional CSV file prefix (default: results_)
• Table selection
  • -t or --table NAME Results table name (default: BENCHMARK_RESULTS)
  • -o or --owner SCHEMA Optional schema owner prefix

Sample Output

The following shows the output of probably the most common use of the command. By default it will fetch the last 3 results from the database to compare.

Colours help call out “hot” percentages (e.g., red for >= 20%, yellow for >= 5%, green otherwise).

CSV Export

Add --csv-dir out to write four files:

• overview.csv – Summary metrics by run.
• dml.csv – Select/Insert/Update/Delete/Commit/Rollback totals by run.
• transactions.csv – Per-transaction counts and percentiles across runs.
• waits.csv – Top-N wait events across runs with “% time / waits / time”.

Whilst I haven’t added all of the metrics to the output I have included a python version of this program in the utils directory which can be modified to included any missing metrics you wish to include.

• 2025 4
• 2024 1
• 2023 5
• 2022 8