this is why at the moment i'm working on some sync systems to enable a DNS with multiple relay servers at different locations distributing connections so that if one or more of them go offline for whatever reason, the other servers still answer, and all of the relays in the cluster sync together so whenever new events go into one of the replicas, they are propagated across the cluster instantly

high-availability, reliable relay hosting. doesn't even have to be that expensive, a VPS at $5/month should be able to serve about 1000 users. i've got later designs in mind to create larger clusters where there is a number of "archival" replicas that maintain a larger database and then smaller ones that have a "layer 2" database that propagates queries to the archival replicas and pulls in older data that fell out of the smaller relay caches so that when there is spikes in demand for older events these are pulled in by the first one to get the query and then when stored, replicated across the "cache relay" databases.

building the deployment system for this will be the third task in my list of things to do, once i have built out all the parts for it.