dtsong/cost-analysis

Cost Analysis

Purpose

Model infrastructure costs at current and projected scale, identify optimization opportunities, and establish cost monitoring with budget alerting. Produces a cost breakdown that enables informed architecture and scaling decisions.

Scope Constraints

Reads infrastructure inventories, pricing documentation, and usage metrics for cost analysis. Does not modify files, provision resources, or access billing APIs or financial systems directly.

Inputs

• Current infrastructure inventory (services, providers, tiers)
• Current usage metrics (requests/day, storage volume, compute hours)
• Growth projections or scaling targets
• Budget constraints or cost reduction goals

Input Sanitization

No user-provided values are used in commands or file paths. All inputs are treated as read-only analysis targets.

Procedure

Progress Checklist

• [ ] Step 1: Inventory infrastructure components
• [ ] Step 2: Estimate per-unit costs at current scale
• [ ] Step 3: Model cost projections at scale
• [ ] Step 4: Identify optimization opportunities
• [ ] Step 5: Design cost monitoring and alerting
• [ ] Step 6: Plan budget allocation and review cadence

Step 1: Inventory Infrastructure Components

Catalog all cost-bearing components:

• Compute: Application servers, serverless functions, background workers, build runners
• Storage: Object storage, block storage, database storage, backup storage
• Database: Managed database instances, read replicas, connection poolers
• CDN: Bandwidth, edge compute, cache storage
• Third-party services: Auth providers, email/SMS, payment processing, analytics, error tracking
• DNS and networking: Domain registration, DNS queries, load balancers, NAT gateways, data transfer
• Email: Transactional email, marketing email, inbound processing

Step 2: Estimate Per-Unit Costs at Current Scale

For each component, calculate:

• Monthly base cost: Fixed costs regardless of usage (reserved instances, minimum tiers)
• Variable cost: Per-request, per-GB, per-user marginal costs
• Cost per user: Total infrastructure cost divided by active users
• Cost per request: Total infrastructure cost divided by total requests
• Document pricing tier thresholds and current utilization against limits

Step 3: Model Cost Projections at Scale

Project costs at growth milestones:

• 2x scale: Which components scale linearly vs step-function? Where do tier upgrades hit?
• 5x scale: Which pricing tiers break? Where do volume discounts apply?
• 10x scale: What architectural changes become necessary? Which components become dominant costs?
• Identify cost cliffs — points where a small usage increase triggers a large cost jump

Step 4: Identify Optimization Opportunities

Evaluate cost reduction strategies:

• Right-sizing: Over-provisioned instances, unused reserved capacity, oversized database tiers
• Reserved/committed use: Savings from 1-year or 3-year commitments on stable workloads
• Spot/preemptible instances: Suitable workloads for interruptible compute (batch jobs, builds)
• Caching to reduce compute: CDN caching, application-level caching, database query caching
• Query optimization: Slow queries consuming excess database resources, missing indexes
• Architecture changes: Serverless for bursty workloads, edge compute for latency, static generation

Step 5: Design Cost Monitoring and Alerting

Establish ongoing cost visibility:

• Budget thresholds: Alert at 50%, 75%, 90%, 100% of monthly budget
• Anomaly detection: Unexpected cost spikes from runaway processes, misconfigured auto-scaling, or attacks
• Cost-per-user trending: Track unit economics over time to catch efficiency degradation
• Tag-based allocation: Cost attribution by service, team, environment, feature
• Review dashboard: Real-time cost breakdown accessible to engineering and leadership

Step 6: Plan Budget Allocation and Review Cadence

Define the financial process:

• Budget allocation: Per-service or per-team budget breakdown
• Review cadence: Monthly cost review meetings, quarterly budget adjustments
• Cost ownership: Which team owns which infrastructure costs
• Approval process: Threshold for new infrastructure spending requiring approval
• Cost-benefit framework: How to evaluate infrastructure investments against engineering time

Compaction resilience: If context was lost during a long session, re-read the Inputs section to reconstruct what system is being analyzed, check the Progress Checklist for completed steps, then resume from the earliest incomplete step.

Output Format

# Cost Analysis: [Project/Service Name]

## Infrastructure Cost Table

| Component | Provider | Tier | Monthly Cost | Cost Driver | Notes |
|-----------|----------|------|-------------|-------------|-------|
| App Server | [provider] | [tier] | $X | requests | ... |
| Database | [provider] | [tier] | $X | storage + queries | ... |
| CDN | [provider] | [tier] | $X | bandwidth | ... |
| **Total** | | | **$X** | | |

**Cost per user**: $X/month | **Cost per 1K requests**: $X

## Scaling Projections

| Component | Current | 2x | 5x | 10x |
|-----------|---------|-----|-----|------|
| Compute | $X | $X | $X | $X |
| Database | $X | $X | $X | $X |
| Storage | $X | $X | $X | $X |
| **Total** | **$X** | **$X** | **$X** | **$X** |

## Optimization Recommendations

| Optimization | Estimated Savings | Effort | Risk | Priority |
|-------------|-------------------|--------|------|----------|
| [description] | $X/month (Y%) | Low/Med/High | Low/Med/High | P1/P2/P3 |

## Budget Alert Thresholds

| Threshold | Monthly Amount | Action |
|-----------|---------------|--------|
| 50% | $X | Review dashboard |
| 75% | $X | Investigate anomalies |
| 90% | $X | Escalate to lead |
| 100% | $X | Freeze non-critical spending |

Handoff

• Hand off to deployment-plan if cost optimization requires changes to deployment strategy (e.g., switching from blue-green to rolling to reduce compute overhead).
• Hand off to observability-design if cost monitoring needs integration with existing alerting and dashboard infrastructure.

Quality Checks

• [ ] All cost-bearing infrastructure components are inventoried
• [ ] Per-unit costs (per user, per request) are calculated for current scale
• [ ] Scaling projections identify cost cliffs and tier boundaries
• [ ] Optimization recommendations include estimated savings and effort
• [ ] Cost monitoring covers budget alerts and anomaly detection
• [ ] Budget review cadence and cost ownership are defined
• [ ] Third-party service costs are included (not just cloud infrastructure)
• [ ] Cost projections account for both linear and step-function scaling

Evolution Notes

<!-- Observations appended after each use -->