ECS metric does not match EC2's

Question

I am currently using AWS ECS to run a dockerized app on EC2 (t3.small instance). When I look at my cluster metrics in the console, it doesn't match at all what I can see on the EC2 instance metrics dashboard.

Here is the CPU utilization from ECS dashboard :
And from EC2 :


What can explain such a difference ? Could it be because of the ECS agent ?

Answer 1

Looking at this a year later I feel very dumb because it actually matches, the time periods are just differents. If you look 10:40 to 11:30 it looks like a perfect match.