you are an sre/devops that owns a web server
this web server can throw errors, and it self reports error rates
the error rates are returned at discrete time steps (for example, every 5 minutes)
you want to know if the server is unhealthy, that way you can stop playing ping pong

a server is unhealthy if the average error rate in a period exceeds some threshold

question:
determine if the server is unhealthy at a particular time

example:
error rates [ 0.1 0.2 0.9 0.9 0.3 0.1 0.01 ]
these are error rates at time 0, 1, etc

for a period of length 2, and time 1, with threshold 0.5, the server is unhealthy
for a period of length 3, and time 6, with threshold 0.3, the server is healthy