Go makes concurrency feel effortless.
Add the go keyword, and you have parallel work happening. It feels powerful. It is powerful.
But power without constraints is how systems collapse.
The Naive Approach: Unbounded Concurrency
While building a small CLI tool to verify large list of domains, my initial implementation spawned one goroutine per domain.
Meaning the number of go routines scaled directly with the input size.
This is a poor design. A large enough input size can cause the system to melt down and crash.
The fix? - Worker Pool
Instead of goroutines growing with input, a bounded worker pool processes domains through a buffered channel. This way, the concurrency level stays constant.
Input size only affects how long the job runs, not how much pressure the system experiences.
That’s the shift: from unpredictable resource consumption to a system that degrades gracefully under load, instead of failing catastrophically.