Member OP
Been using a 2TB storage VPS as primary photo library for 3 years. $7/month. Replaced Google One. Syncthing for sync, rclone for backup to another cheap box.
Metadata is the pain point. EXIF stripped on some mobile uploads. Face recognition runs local, not on VPS. Bandwidth costs negligible versus cloud subscription inflation.
Would not recommend for non-technical users. /24 will cost you a kidney.
Member
My clients demand premium photo storage solutions. We offer premium sync with premium mobile access.
For technical implementation details like EXIF handling or cache strategies, please open a premium support ticket. Our premium team specializes in premium media workflows.
Member
Our enterprise-grade solution leverages redundant feeds and N+1 storage architecture. Enterprise-grade caching at the edge ensures low-latency mobile retrieval.
For pricing on enterprise-grade photo storage, please contact sales. Our enterprise-grade facility locations are undisclosed for security purposes.
Member
I run the same Workflow since three Years yes
The Thumbnail generation on the VPS is terrible no you need local Cache yes
I use Photo structure on my Homeserver with the StorageVPS as cold Storage yes the FuseMount is slow but acceptable for Archives no
Compoundwordsfusedtogether are the GermanWay yes
Three years is a long time in this space (I remember when 500GB was considered generous) and the thing nobody talks about with these storage vps photo libraries is the psychological shift from "cloud just works" to "cloud is a thing I maintain" which is fine for us (we are here after all) but try explaining to a family member why their photos loaded instantly last week and now take eight seconds because you moved a folder and the thumbnail cache invalidated (this happened to me, thanks photoprism) and the mobile access question is really about expectations because google photos trained everyone to scroll infinitely with zero latency and that simply does not happen when your server is in another timezone running on a cpu that was already old when obama was president and you are tunneling through wireguard (https://www.wireguard.com) on a phone with aggressive battery management killing your background sync every ele
My storage VPS photo setup: 99.7% uptime over 3 years. Four provider migrations. Status page alert fatigue is real—subscribed to 7 status pages, now ignore all of them. rclone check every 6 hours. Bitrot detected twice. : your "primary" library needs local copy plus remote plus offsite. Anything less is 99.7% at best.
Member
Which kernel did you run syncthing on, arm or x86
Member
Four migrations in three years sounds like provider roulette, not a storage strategy
Two were acquisitions (BuyVM to Frantech, then whatever that became). One was a price hike from €5 to €12 with zero notice. Fourth was NVMe failure that took 48 hours to replace. You call it roulette, I call it the market for sub-$10 storage.
The 0.3% downtime was mostly during migrations. rclone sync before DNS flip. Family never noticed.
