A long-running, stateful service runs on a single xlarge general purpose On-Demand EC2 instance. Metrics show it uses about 80% of memory and 40% of CPU. The SysOps administrator needs to lower costs without degrading performance. Which instance type change meets this goal?

The service is memory-intensive (80% memory utilization) and moderately CPU-intensive (40% CPU utilization). Switching to a large memory-optimized instance (e.g., R-series) provides more memory per vCPU and lower cost than an xlarge general purpose instance, aligning with the goal of lowering costs without degrading performance. A large instance is generally cheaper than an xlarge instance. Switching to a compute-optimized instance would reduce memory, potentially degrading performance. Switching to a Spot instance introduces interruption risk, which is unsuitable for a "long-running, stateful service." Switching to two large general purpose instances would increase cost due to running two instances instead of one, and also introduces complexity for a stateful service.