Database And Compute Performance Scaling — AWS SysOps Administrator (SOA-C03)
Read Replicas Fix Reads, Not Write Contention
A scenario describes RDS performance degradation under high load. The instinct is to add a read replica. If the bottleneck is read-heavy reporting queries, that's correct. If the bottleneck is write throughput or connection exhaustion, a read replica provides no relief. The exam distinguishes between read capacity scaling and write scaling — they require different patterns, different services, and different configuration choices. Identify the bottleneck type before selecting the scaling approach.
What This Pattern Tests
The exam describes a performance bottleneck and tests the right scaling technique. Read replicas (RDS, Aurora up to 15 replicas) handle read-heavy workloads by distributing SELECT queries. ElastiCache Redis/Memcached handles repeated queries by serving results from memory (microsecond reads vs. millisecond database reads). DynamoDB DAX provides microsecond reads for DynamoDB-specific workloads. Write scaling requires partitioning/sharding (DynamoDB handles this automatically with partition keys). The trap is adding read replicas for a write-bottleneck (replicas only handle reads) or caching for data that changes every second (cache invalidation becomes the bottleneck).
Decision Axis
Bottleneck type determines technique: read-heavy = replicas, repeated queries = cache, write-heavy = partitioning, compute-heavy = scale out.
Associated Traps
More Top Traps on This Exam
Decision Rules
When repeated identical queries dominate read traffic and the acceptable staleness window is longer than cache TTL, an in-memory cache layer eliminates query execution entirely and resolves CPU saturation; a read replica only redistributes the same full query executions across more compute without achieving cache-hit-ratio gains.
When read requests are highly repetitive and data volatility is low, an in-memory cache absorbs load at lower latency and zero database CPU cost; a read replica is preferred only when reads are genuinely divergent and cannot be served from a shared cached result.
Domain Coverage
Difficulty Breakdown