Does the peering between two transit gateways is a AWS managed connection ?

we have a big mesh of transit gateways that is distributed on different regions and some of them even exists in the same region connecting different accounts.

my question is that, does the peering between two transit gateways is AWS managed connection so it will not be go down eventually or I need to make a redundant path to connect VPCs behind these two connecting transit gateways

I am asking this question because the peering between transit gateways does not ask for speed or bandwidth for example when create the peering.