Checkout speed in own gitlab runner

Why are there so much differences in checkout speed?

I have own runners in my k8s cluster which i use with, this worked super well and the build times were as i would expect … until several months. Now the checkouts take at least 10 minutes and sometimes it even timeout.

To verify that this is not an issue in my kubernetes cluster, i checked out various repositories from different locations inside the gitlab runner pod.

starts good 7mb/s and immedialty goes down to 1.000 kb/s - 500 kb/s

starts good 7mb/s and goes up to max 10mb/s, swining at top 10 / 11mb / s

starts good, and goes up to max 11 mb/s

swinging between 5-10mb/s

i can reproduce these values pretty constantly. So it looks like either gitlab hosts some repositories only on slow servers or servers far away from my location (germany). Is there any chance to influence that, e.g. by buying premium? I really do not want to manage a gitlab instance but with that slow checkouts there is something i have to do.

Thank you and best regards