EC2 instances carry most AWS-hosted workloads, from web tiers to batch processors. Teams plan capacity around the assumption that an instance will be there when needed, until a noisy neighbor, an exhausted disk, or an unexpected reboot proves otherwise.

Because the Gremlin Agent installs directly on the instance, you get full host coverage. Run a CPU or Memory experiment to see how your application handles resource pressure, a Disk or IO experiment to test a slow or full volume, or a Shutdown experiment to simulate the instance disappearing outright.

Gremlin's built-in reliability tests turn these one-off checks into a repeatable practice. The CPU, Memory, and Disk I/O tests validate your autoscaling thresholds, and the Host and Zone tests confirm your service survives losing an instance or an entire availability zone.