To achieve the best performance from our instances, always consider the data access pattern, the throughput required to the database, and the memory requirement for your database servers; this will allow you to select the appropriate instance type. Once in production, the instance should be monitored for I/O and latency, and of course, disk and memory usage. We can monitor our instances using CloudWatch, and we can also set up alarms on CloudWatch to determine when our instance health or resource usage is spiking and an action would need to be taken to resolve the issue. There are four ways to increase the performance of our clusters:

• Change the instance type to increase the performance of the instance
• Add read replicas to distribute the read load over more instances
• Shard the data to multiple instances
• Add caching to offload the read operations to the cache (discussed in Chapter 12, Introduction to ElastiCache)