I’m about to rebuild a server cluster - currently set up for high reliability & availability - RAID, disk mirroring, automatic failover, all that good stuff.
I’ve been thinking about replacing the storage “plumbing” with a distributed filesystem - that extends beyond the cluster to another location, and to various laptops & desktops - one big file system for an organization & its community.
From 50,000 feet, just turning all the machines into one big IPFS cluster would seem to make sense - but one wonders about all the processing & data transfer overhead. Also, the latency - how fast will systems sync if they’re on neighboring machines. And then there's the matter of checkpointing & reliable failover.
Can anybody point at some analysis of the processing & network overhead associated with IPFS & IPFS-cluster? Any experience and/or suggestions for using IPFS as a generic, distributed, high-availability storage network? Case studies? Places to start looking? (Note: Didn't get a single response on discuss.ipfs.tech - doesn't seem to be very active as a support community.)
Thanks Very Much,
Miles Fidelman
