Hi there,
this is Matthias, from the german OSM community. Some of you might already know, that we think about to start a panoramax server for DACH. In a closed Beta, we will try to find out, what the technical and workflow / legal requirements are and how they will work here in DE.
But in the long run we try to find a way for a sustainable hosting solution. While I personally like the idea of selfhosting and distributed nodes, I’m not sure if this will work for such a storage / traffic consuming platform like panoramax?
Thats why I like to ask, what is your resume in hosting a EU wide service? What would be a good technical approach to add more and more disks and how about a hosting location? How do you manage service requests like bulk imports / exports?
Cloud based object storage are way too expensive (around 8€/TB/month)
Renting bare metal servers are cheaper (1.5 to 4€/TB/month) but quickly reach the same cost, as you need to deal with backups and/or redundancy.
On premise is much cheaper (10€/TB for used HDDs but only once), but you need to invest (step by step) in equipment, set it up, find a hosting place, administer the hardware/software. That’s the choice made by OSM-FR, it was the only option that was financially possible.
Distributed storage (like Ceph) is something that may become necessary for a single large instance, when one single server can not handle all the storage by itself.
Distributed storage on many self/home hosted nodes (like Garage) seems too complex to deal with because you need even more redundancy, as you can’t be sure all nodes will be available 24/7 or in the long term.
I think we should avoid to setup larges instance, like a EU wide one. This creates much more problems than maintaning for example one instance per country or even smaller area.
Decentralization is a key to the success of Panoramax and we should avoid to mimics “platforms”.