Skip to content
Documentation

Performance Advisor

Three live-data tools fed by your active connection: a linear capacity forecast, an index hotness ranking, and a connection pool size suggester. No simulation — everything reads your actual workload.

What it is

A workspace tab with three tabs: Capacity (forecastLinear over a metric history), Index Hotness (rankIndexHotness using IndexStats), and Pool Tuner (suggestPoolSize from current op concurrency). Each tab has a "Fetch from active connection" button that pulls the live data; you can also paste a JSON sample to test without a live connection.

When to use

Run Capacity before a major release to forecast when storage / RU / IOPS will saturate. Run Index Hotness once a quarter to find indexes nobody touched in 90 days. Run Pool Tuner when application logs show MongoDB connection wait — it tells you the optimal pool size for the workload you actually have.

How to open

  1. In the NoSqlStudio desktop app, open the Tools menu → Performance Advisor, or press Ctrl+Alt+Shift+F.
  2. A workspace tab opens with three tabs at the top: Capacity / Hotness / Pool.
  3. Each tab has a "Fetch from active connection" button. Click it to pull live data. The output renders as stats cards and (where applicable) a chart.

Each pane

  • Capacity — pulls 14 days of metric history (or whatever range you set), runs forecastLinear, and projects when the metric crosses a configurable threshold. Useful for storage growth, RU/sec saturation, IOPS limits.
  • Index Hotness — pulls $indexStats from every collection in the active database, ranks indexes by accesses-per-day, and flags indexes touched 0 times in the lookback window. Drop candidates appear in red.
  • Pool Tuner — pulls current op concurrency, applies a Little's-Law-style suggestion based on observed wait times, and outputs the recommended poolSize for your driver config.

Limitations (v1)

  • Forecast is linear — does not model seasonality or piecewise regimes. For seasonal workloads (e.g. day/night cycles) interpret the slope as a worst-case ceiling, not an exact projection.
  • Index Hotness reads $indexStats, which resets on every mongod restart. If your cluster restarted yesterday the 90-day window is empty — wait for stats to accumulate or extend the window.
  • Pool Tuner assumes your driver is the bottleneck (not server CPU / replica election). On a stressed cluster the recommendation may oscillate.