MQTT Performance Visualization Tools

A toolkit for visualizing and analyzing MQTT connection, messaging performance, and latency data.

Overview

This project provides scripts for generating various visualizations of MQTT performance data:

• Connection latencies (TCP, TLS handshake, MQTT Connect)
• Publishing latencies
• Success rates for various operations
• Latency distributions and trends over time

Requirements

Python Dependencies

The scripts require Python 3.6+ and the following packages:

pip install pandas numpy matplotlib

Or install using the requirements file:

pip install -r requirements.txt

Scripts

Main Script

• bin/batch.sh: Batch processing script that generates all visualization types for a given log directory

Individual Visualization Scripts

• bin/plot_qoe.py: Generate box plots for QoE metrics (latency distributions)
• bin/plot_time2.py: Create time-series plots showing latency trends
• bin/plot_bar_qoe.py: Create bar charts showing latency distributions
• bin/plot_rate.py: Generate plots showing success rates and cumulative operations
• bin/plot_time.py: Alternative time-series plot script

Usage

Process a Single Log Directory

bin/batch.sh /path/to/log/directory

Process Multiple Log Directories

bin/batch.sh
# This will process all directories under ~/Home/tmp/toy_bench_data

Individual Metrics

Publishers MQTT.CONNECT Latency

bin/plot_qoe.py 'TCP,Handshake,Connect' /path/to/publog-*.csv

Subscribers Latency and Publish E2E Latency

bin/plot_qoe.py Publish /path/to/sublog-*.csv

Publish E2E Latency Time-series

bin/plot_time2.py Publish /path/to/sublog-*.csv

Connect E2E Latency Time-series

bin/plot_time2.py Connect /path/to/publog-*.csv

Publish E2E Latency Distribution Bar Chart

bin/plot_bar_qoe.py Publish /path/to/sublog-*.csv

Connect E2E Latency Distribution Bar Chart

bin/plot_bar_qoe.py Connect /path/to/*.csv

Operation Rate Visualization

bin/plot_rate.py Handshake /path/to/*.csv

CSV Format

The CSV files used by these scripts are generated by emqtt_bench, a benchmarking tool for MQTT brokers.

Supported CSV Headers

The scripts support analyzing CSV files with the following format:

Publisher Log Files (publog-*.csv)

ClientId,TS,TCP,Handshake,Connect[,Subscribe]
client1,1616168562000,35,120,50,-1
client2,1616168562500,42,115,48,-1

Subscriber Log Files (sublog-*.csv)

ClientId,TS,TCP,Handshake,Connect,Subscribe,Publish
sub1,1616168562000,30,118,45,25,73
sub2,1616168562500,38,125,52,28,65

Field Explanations

• ClientId: Unique identifier for the MQTT client
• TS: Timestamp in milliseconds when the operation was initiated (epoch time)
• TCP: Time in milliseconds taken to establish the TCP connection
  • Measured from socket creation to successful connection
• Handshake: Time in milliseconds for TLS handshake completion
  • Measured from from socket creation to successful completion
  • Includes certificate validation and cipher negotiation
• Connect: Time in milliseconds for MQTT CONNECT operation
  • Measured from socket creation to receiving CONNACK
• Subscribe: Time in milliseconds for MQTT SUBSCRIBE operation
  • Measured from from socket creation to receiving SUBACK
  • -1 indicates operation didn't occur or failed
• Publish: End-to-end latency in milliseconds for message delivery
  • Measured from message publication to message receipt by subscriber
  • -1 indicates operation didn't occur or failed

Note: The -1 values in the data are for compatibility with legacy data formats. Newer formats use NaN instead of -1 to represent missing or failed operations. The scripts handle both representations by converting them to NaN values during processing.

Generated Output

Scripts generate PNG files with naming patterns like:

• {testname}_{operation}_bar_qoe.png: Latency distribution bar charts
• {testname}_{operation}_latency_trends.png: Time-series latency trends
• {testname}_{operation}_rate.png: Operation rate charts
• {testname}_{fields}_qoe.png: QoE distribution box plots