We've been working quite a bit of our NFS storage topology and have recently introduced some remediation to prevent outages based on NFS availability. Previously an NFS failure would have pretty wide-reaching implications, now it's a lot more isolated.
We're also working on an entirely new storage architecture for scaling and are slowly rolling this out. You can see more here: https://gitlab.com/gitlab-org/gitaly
We're also working on an entirely new storage architecture for scaling and are slowly rolling this out. You can see more here: https://gitlab.com/gitlab-org/gitaly