In my project, file synchronization process, involves file transfer to 16 servers, going forward we will be doing the same on more than 1000 servers. I would like to know if SyncThing is capable of handling transfer to large number of servers. If yes, will there be any slowness due to large number of servers. If yes, what will be the best an alternative solution?
Why not? Where do you see limits? Can your OS’es tcp/ip stack handle more than a thousand established tcp connections or more than a thousand udp streams?
Simple solution would be for me, dont peer all nodes with all others. Create groups and hierarchy of some sort that makes sense.
Maybe most intelligently according to your lan and wan setups and topology.
Syncthing keeps the index for each connected device in memory (iirc), so using a full mesh with 1000 servers probably isn’t a good idea. So grouping like @usernamegoeshere suggested is the way to go.
No, mostly on disk, but there is some memory overhead per connection indeed.