Thank you so much! Yes, async can be acceptable as long as we can handle any failure properly. I am asking here because I did not see anything covering multi-site in lakefs
docs.In terms of the limited line, We already have a multi-site pipeline, the bandwidth is enough for our current cross site data flow. However, If we can have an uniformed view over sites by leverage lakefs,it would be excellent! In our concerns, It is more important regarding the fault tolerance. How can we handle the network partition or one site is completely unavailable. And the metadata should not be corrupted or should be at least fixable once network is back to normal.