if everyone on this corner of the fediverse put all of our self-hosted server workloads into one collectively operated high availability cluster with bin packing and all the goodies I wonder how much excess capacity we'd find that we had available, even accounting for the redundancy necessary to achieve a true HA state for all our services
I mean, we are already sharing compute resources between our fediverse instances with little to no accountability as to quotas or resource management between us. if we formalized and generalized the setup, even in small pockets, there could be some very cool knock-on effects I'd suspect.
there's part of my brain that says that the federated distributed MMO that @cwebber is working toward (and beyond) and the underlying tech will also add some interesting dynamics into the possibilities here - what happens when we start to share secure object capabilities with each other that allow for some subset of arbitrary computation rather than something much more limited in application and scope?
and, like, it doesn't matter really how small or large the individual HA clusters are above some minimum threshold for fault tolerance. I'd wager a thousand pocket clusters are as good as ten pocket clusters each a hundred times the size when your talking about multi-operator heterogeneous workload allocation.
@djsundog i am at this time reminded of... there was some collection of fedi servers that decided to go in together on image hosting, i thought jorts was involved but now i can't find who. I wonder what happened to that.
@djsundog i bet they would have some helpful Lessons Learned ™️ to share forward on yr project concepts!
The social network of the future: No ads, no corporate surveillance, ethical design, and decentralization! Own your data with Mastodon!