An educational technology company is transitioning a legacy grading system from its internal data center to Amazon EC2 instances. As part of the migration, the cloud architecture team must implement infrastructure monitoring. One requirement is to avoid reacting to brief CPU usage spikes above 50%, which are common during batch uploads. However, if the CPU usage exceeds 50% and disk read IOPS are simultaneously high, the team must be alerted immediately to prevent performance degradation. Reducing false alarms is a priority. What is the most effective way to meet these needs?