Building LogSum: A 3.3ms Log Analyzer with a Beautiful Terminal UI

Mustafa Yildiz — Fri, 13 Jun 2025 16:45:05 +0000

I've just released LogSum v0.1.0, a fast log analyzer that can process 10,000 log entries in just 3.3 milliseconds. But speed wasn't the only goal - I wanted to make log analysis actually enjoyable.

The Problem

We've all been there: production issues at 3 AM, scrolling through endless logs, trying to spot patterns with tired eyes. Existing tools either:

Dump everything to stdout (grep, awk)
Require complex query languages (LogQL, KQL)
Need heavy infrastructure (ELK stack)
Cost a fortune (Datadog, Splunk)

I wanted something that just works, runs locally, and makes patterns obvious.

The Solution: LogSum

LogSum combines three things that rarely go together:

Blazing fast performance - 3.3ms for 10K entries
Beautiful terminal UI - with animations and colors
Intelligent analysis - automatic pattern detection

Key Features

🔍 Smart Pattern Detection

$ logsum analyze app.log
✓ Analyzed 10,000 entries in 3.3ms
✓ Found 7 patterns
✓ Generated 3 insights

LogSum tries to identify:

Error patterns and anomalies
Performance degradations
Status code distributions
Timeline-based spikes

🎨 Interactive Terminal UI
The TUI features:

Rainbow animated borders (because why not?)
Keyboard navigation between patterns/insights/logs
Real-time pattern highlighting
ASCII charts for visualizations

📊 Multiple Output Formats

# Beautiful terminal output (default)
logsum analyze app.log

# Machine-readable JSON
logsum analyze app.log -f json

# Markdown report
logsum analyze app.log -f markdown -o report.md

# CSV for further analysis
logsum analyze app.log -f csv -o patterns.csv

Technical Implementation

Performance Secrets

The 3.3ms performance comes from:

Zero-copy parsing where possible
Efficient pattern matching
Efficient data structures - avoiding allocations in hot paths

Architecture

LogSum follows clean architecture principles:

internal/
├── parser/      # Multi-format log parsing
├── analyzer/    # Pattern detection engine
├── formatter/   # Output formatting
└── ui/          # Bubble Tea TUI

Each component is independent and testable. The parser handles JSON, logfmt, and plain text logs automatically.

The TUI Magic

Built with Bubble Tea, the UI updates at 30 FPS for smooth animations:

type model struct {
    activeView  View
    patterns    []Pattern
    insights    []Insight
    borderColor int // for rainbow animation
}

Try It Yourself

# Install
go install github.com/yildizm/LogSum/cmd/logsum@latest

# Analyze your logs
logsum analyze /var/log/app.log

# Watch mode for real-time analysis
logsum watch /var/log/app.log

# Get sample logs to try
curl -L https://github.com/yildizm/LogSum/raw/main/testdata/sample.log > sample.log
logsum analyze sample.log -v

What's Next?

This is just v0.1.0! New features are coming soon.

PS

The project is open source and MIT licensed. I'd love your feedback, bug reports, and contributions!

Links:

GitHub: github.com/yildizm/LogSum
Issues: github.com/yildizm/LogSum/issues

What log analysis features would you like to see? Let me know in the comments!

DEV Community: Mustafa Yildiz