Third AWS SimuLearn course, Computing Solutions. The scenario is a school's class scheduling system, running on an EC2 instance that's outgrown its own compute and memory. The fix ends up being straightforward, resize the instance. What's actually worth writing about is a detail sitting quietly in the middle of the architecture diagram: two completely different ways to reach the same instance, both leading to the exact same box.
Two paths to the same instance, and why that's not redundant description, it's a real choice
The diagram shows a terminal connecting through EC2 Instance Connect on one path, and AWS Systems Manager on a separate path, both terminating at the same t3.micro instance. That's not the diagram being sloppy about drawing one connection twice, it's showing two genuinely different mechanisms for the same job, and the difference between them matters more than it looks like at a glance.
EC2 Instance Connect is closer to what most people picture when they think "SSH into a server," a terminal-based connection to the instance directly. AWS Systems Manager's Session Manager gets you the same practical outcome, a shell on the instance, without needing an open inbound port for it. No security group rule allowing SSH traffic in, no bastion host to manage, no key pair to lose track of. The connection is brokered through the Systems Manager agent already running on the instance, initiated from the AWS side rather than requiring the instance to accept an inbound connection at all.
For a school's scheduling system, the practical version of this is simple: every open port is one more thing that has to be locked down correctly and kept that way. Session Manager removes an entire category of "did we configure this security group right" from the list of things that can go wrong, not by making SSH safer, but by not needing it in the first place.
Resizing is the visible part, choosing the right instance type is the actual decision
The scenario's fix is moving from t3.micro to m4.large, and the objectives frame this as comparing instance families and types, not just "picking a bigger number." That distinction is worth taking seriously. A t3 instance is burstable, it runs on a baseline level of CPU performance and accumulates credits it can spend on short bursts above that baseline, which is a good fit for workloads that are mostly idle with occasional spikes, exactly what a small pilot version of a scheduling system probably looked like at first. An m4 instance doesn't work on that credit system, it's built for more sustained, predictable compute and memory demand.
So the real decision isn't "t3.micro was too small, make it bigger." It's "this workload's actual usage pattern outgrew what a burstable instance is designed for in the first place," which is a workload-shape question, not just a size question. Picking a bigger t3 would have addressed the size problem and left the shape mismatch untouched.
What the lab actually covered
Filtering and identifying instance types by workload attributes, connecting via Session Manager and reading an instance's own metadata to confirm what's actually running before changing anything, and starting and stopping an instance through the console as part of resizing it, since an EC2 instance type change happens while the instance is stopped, not live. The DIY step, changing the instance type to m4.large, is the resize itself, the part that looks like the whole exercise from the outside but is really just executing a decision made earlier in the process.
Loading article