Scalability Assessor

1. Can the application run multiple instances behind a load balancer?
2. Are there shared resources that become contention points under load?
3. Assess database connection scaling: connection pooling, read replicas.
4. Check for distributed locking needs and their implementation.
5. Evaluate message queue consumer scaling patterns.

Step 3 — Vertical Scaling Limits

1. Identify CPU-bound operations that benefit from more cores.
2. Find memory-bound operations that benefit from more RAM.
3. Assess I/O-bound operations limited by disk or network throughput.
4. Determine the single-machine ceiling for each resource type.
5. Estimate cost curve for vertical scaling [INFERENCIA].

Step 4 — Scalability Report

1. Classify overall scalability: linear, sub-linear, or non-scalable with evidence.
2. Identify the scaling bottleneck (the first thing that breaks under load).
3. Recommend a scaling strategy: horizontal, vertical, or hybrid.
4. Provide a capacity planning estimate for target load levels.

Quality Criteria

• Statefulness assessed per component, not globally [HECHO]
• Scaling bottleneck identified with specific evidence
• Both horizontal and vertical scaling evaluated
• Capacity estimates include assumptions clearly stated [SUPUESTO]

Anti-Patterns

• Assuming statelessness without checking for in-memory state
• Planning only for horizontal scaling when the database is the bottleneck
• Ignoring cost curves (10x load may not need 10x resources)
• Treating scalability as purely a infrastructure problem (often requires code changes)