capacity-planner

name	capacity-planner
description	Analyzes database resource utilization (CPU, memory, disk, connections) and forecasts growth to plan capacity upgrades before performance degrades.
allowed-tools	Read, Write, Bash
effort	medium

Capacity Planner

When to activate

When database resource utilization exceeds 70% on any metric, when planning for expected traffic growth, before seasonal peaks (Black Friday, product launches), or when storage is projected to fill within 90 days.

When NOT to use

Skip for databases running under 40% utilization with no expected growth change. Skip if auto-scaling is already configured and tested.

Instructions

1. Collect current metrics: CPU avg/p95, memory usage, disk I/O, connection count, storage used/total, replication lag.
2. Identify growth trends: weekly storage growth rate, monthly connection growth, seasonal traffic patterns.
3. Forecast time-to-capacity for each resource at current growth rate.
4. Flag any resource projected to hit 85% within 90 days.
5. Recommend actions: vertical scaling (instance upgrade), horizontal scaling (read replicas, sharding), query optimization, archival/cleanup, or connection pooling.
6. Estimate cost impact of each recommendation.

Example

Scenario: PostgreSQL on 8 vCPU / 32GB, 400GB of 500GB used.

CAPACITY REPORT: prod-db-01

Current (30-day avg):
  CPU: 62% avg / 89% p95          ⚠️
  Memory: 78% (25GB/32GB)         ⚠️
  Disk: 400GB/500GB (80%)         ⚠️
  Connections: 180/200 max         ⚠️
  Replication lag: <1s             ✅

Forecast:
  Disk at 85% in: 18 days (growing 2.1GB/day)
  Connections at 85% in: 34 days
  CPU p95 >90% in: 12 days

Recommendations:
  1. IMMEDIATE: Add PgBouncer connection pooler (reduces connections 60%)
  2. THIS WEEK: Archive orders older than 2 years to cold storage (-80GB)
  3. THIS MONTH: Upgrade to 16 vCPU / 64GB instance
  4. Add read replica for analytics queries (offloads 30% CPU)