Post
Not sure, but to put this differently, I think @Sascha is asking if #ATProto has anything similar to nomadic identity, provided by Mike Macgirvin’s Hubzilla, (streams), and Forte.
https://medium.com/@tamanning/nomadic-identity-brought-to-you-by-hubzilla-67eadce13c3b
(continues)
So... on a second Hubzilla instance B, a user can create a clone of her channel on instance A. The clone includes everything—posts, connections, articles, wikis, etc. From then on, any activity the user does on either A or B—new posts, connections, etc—get mirrored on the copy.
Of course, the user can also create additional clones on instances C, D, etc.
(continues)
@ Feral Thoughts @ Sascha Foerster :bonndigital:
With nomadic identity, your identity is behind your channel. You clone your channel, and if one of the clones fails, you continue working with the next one.
However, this means that your data is stored on multiple servers, which are essentially just "mirrors".
The AT Proto PDS is not mirrored on multiple servers and does not clone its data to another server; the PDS is the server itself and is therefore independent of whether any AT Proto access or application fails.
The idea behind AT Proto is that your data is not stored on any server other than your own server (PDS). You could also say that the PDS is like a trailer that you attach to different cars. If one car fails, you attach the trailer to the next car. However, the car has none of the contents of your trailer.
So... I guess @Sascha is asking if such an arrangement is possible in case of #ATProto, either at the level of the PDS, or at some other level.
Can a user simultaneously have more than one “official” copy of their data (posts + connections + blogs + ...), with all the clones mirroring new activites on any one of them? And if one of them shuts down or gets blocked, will the others continue functioning uninterrupted?
If not in practice today, is this possible in theory?
@feralthoughts @Sascha Hmm… so yeah, in theory, you probably could set up such system where you have repos on multiple hosts which are syncing automatically from a "master", and if the active one becomes dead, then you switch over to make one of the others become the active one. But the "failover" wouldn't be automatic, you'd need to manually or through some code make a write operation to the DID document, changing the assignment to a different host.
@feralthoughts @mackuba Thanks both of you, I learned a lot from your posts and links. Now I need some time to let that sink in. :)
A space for Bonfire maintainers and contributors to communicate
