Database Scaling Strategy Advisor

You are a database architect who has scaled systems from thousands to billions of records. Help me design a database scaling strategy.

Current database: [CURRENT_DB]
Current data volume: [DATA_VOLUME]
Growth rate: [GROWTH_RATE]
Read/write ratio: [READ_WRITE_RATIO]
Query patterns: [QUERY_PATTERNS]
Latency requirements: [LATENCY_REQUIREMENTS]
Consistency requirements: [CONSISTENCY]
Budget: [BUDGET]

Design a scaling strategy:

**1. Current State Assessment**
- Bottleneck analysis: Is the problem reads, writes, storage, or connections?
- Query performance audit: N+1s, missing indexes, full table scans
- Connection pool analysis

**2. Vertical Scaling** (do this first):
- Hardware upgrades (CPU, memory, IOPS)
- Configuration tuning (buffer pool, work_mem, max_connections)
- Query optimization (indexes, materialized views, query rewriting)
- When you've hit the ceiling and need to go horizontal

**3. Read Scaling**
- Read replicas: How many, routing strategy, replication lag handling
- Caching layers:
  - Application-level cache (what to cache, TTL strategy, invalidation)
  - Redis/Memcached: Cache-aside vs. read-through vs. write-through
  - CDN caching for API responses
- Materialized views for complex aggregations

**4. Write Scaling**
- Write-ahead optimization
- Batch/bulk operations
- Async writes (queue → DB)
- Sharding strategy (if needed):
  - Shard key selection (the most critical decision)
  - Hash-based vs. range-based partitioning
  - Cross-shard query handling
  - Resharding plan

**5. Polyglot Persistence** (right tool for each job):
- Transactional data → [recommendation]
- Search/full-text → [recommendation]
- Caching/sessions → [recommendation]
- Analytics/OLAP → [recommendation]
- Time-series → [recommendation]
- Graph relationships → [recommendation]
- Synchronization strategy between stores

**6. Data Architecture**
- Hot/warm/cold data tiering
- Archival and purge strategy
- Backup and point-in-time recovery

**7. Estimated Cost & Timeline**
- Phase 1 quick wins (this week)
- Phase 2 medium-term (this quarter)
- Phase 3 long-term (if growth continues)
- Cost comparison of approaches

**8. What NOT To Do**
- Premature optimization warnings
- Over-engineering signals
- When managed services beat self-hosting