1. edit server.go, add profiling to the import:

import _ "net/http/pprof"

2. run the server

3. check out the page

http://localhost:8000/count?n=10000000000

4. check out profiling page: http://localhost:8000/debug/pprof/

5. Find the bottleneck with the go tool:

$ go tool pprof http://localhost:8000/debug/pprof/profile?seconds=30s