I've built Dimension-UI, a desktop application for interactive analysis of time-series data, written in Java 25+ with a Swing UI.

While Grafana is fantastic for high-level dashboards, I always found it cumbersome for deep, iterative analysis. The main pain point is its "stateless" drill-down: you select a time range, and you jump to a new, filtered view, losing the original context. To compare different views, you end up juggling multiple browser tabs.

Dimension-UI's core concept is "stateful drill-down." When you select a time range on a chart, a detailed breakdown (like a Gantt chart, pivot table, or raw data) appears directly below it. The main context is never lost, which dramatically reduces cognitive load when hunting for the root cause of an issue.

Key differentiators from a standard Grafana/Prometheus stack:

- High Interactivity: Being a desktop app allows for a much more responsive, stateful interface that's hard to achieve on the web. In my tests, Dimension-UI uses significantly less RAM (~400MB for 300+ charts) compared to a browser tab with a complex Grafana dashboard (~900MB).

- Simplified Stack: It can connect directly to databases (PostgreSQL, Oracle, etc.) via JDBC or to Prometheus via HTTP. For many use cases, this eliminates the need for separate exporters.

- Built-in Advanced Analysis: It includes out-of-the-box anomaly/motif detection using Matrix Profile and forecasting with ARIMA, accessible directly from the UI.

- "No-code" DB Explorer: An Ad-hoc mode lets you point the tool at a database table, select the timestamp and metric columns, and start visualizing and analyzing data without writing SQL.

I wrote a detailed comparison with Grafana, focusing on the task of analyzing Active Session History (ASH) in a PostgreSQL database. It clearly demonstrates the difference in workflow and how much faster you can pinpoint performance issues.

This project was also the first real-world application for Dimension-DI https://news.ycombinator.com/item?id=45821919

Migrating from Dagger 2 to my own DI solution made the codebase much more modular and testable.

The project is open source, and I'd love to get your feedback.

Sample profiles and GIF screencasts are in the repo.

Links:

Dimension-UI (GitHub): https://github.com/akardapolov/dimension-ui

Article: Grafana vs. Dimension-UI for ASH Monitoring: https://habr.com/ru/articles/865622/ (The article is in Russian, but browsers can translate it, and the screencasts are self-explanatory)

Dimension-DI (the DI container it uses): https://github.com/akardapolov/dimension-di