Cloud backup is the future, but not not for everyone — yet
The biggest problem we face with cloud services is upload bandwidth. Most internet services are extremely asymmetric with the model being that we download much more data than we upload — that model really doesn’t fit with photographers trying to use an online backup service where we might want to update lots of GB daily.
Cloud storage isn’t a backup
Don’t confuse Cloud storage services (e.g. Dropbox, OneDrive, iCloud, etc.) with an online/cloud backup service — they are distinctly different beasts. The cloud storage services replicate/copy a set of files from your local disk to cloud storage (and typically sync or make them accessible on other devices too). Very handy, but as I’ve previously discussed this is just as likely to replicate anything that goes wrong (except for a catastrophic disk failure).
Practicalities of Internet/cloud solutions
Let’s play with some numbers.
1Mbit/s translates to approximately 10s/MB. So 1GB is going to take 10000s to upload — roughly 2 3⁄4 hours, but let’s round that to a more conservative 3 hours. With modern image sizes, 1TB of storage doesn’t take long to fill, but if you need to upload at 1MBit/s then your first backup that could take 125 days (with the machine running 24x7).
So your initial backup likely to take weeks or months to complete — my first 1.5TB took close to two months (I started at 2.5Mbits/s up, and was upgraded to 5Mbits/s midway through). Some services will offer to seed your backup by sending them a hard disk (nobody offered this to UK users the last time I looked).
Now we have fibre internet services many readers will laugh at the concept of a mere 1Mbit/s upload, but there will many be others sitting wishing they had that upload speed.
Another consideration is to have the original seed backup complete before adding more data — there is no point in a backup that you can never consider to be complete. This can be a major challenge if your backup is running on your live storage.
Once you have completed your first “seed” backup, simple practicality kicks in — if your typical shoot (or retouching) generates more data than you can upload before you add your next shoot then an online solution just isn’t feasible. it’s not too hard to fill a 16GB card — that could take two days to upload at 1Mbit/s.
Avoid moving files and folders
Modifying your storage structure is also a major consideration — backup software generally isn’t smart enough to understand moving files and folders (it simply sees content removed from one location and new content in a second location).
If you outgrow your local disk storage and move everything to a new (larger) disk consider swapping your new disk to replace your old disk and preserve the same storage path. For Windows users, don’t just move all your files from your old E: drive to your nice new F: drive — swap the drive letters so the files and folders haven’t moved. For Mac users, it’s similar — copy the files then swap the volume names so the storage paths remain the same.
Of course you can’t always avoid having to do this, but just be aware of the ramifications.
Backups record changes
That sounds so obvious — if a file is modified the backup agent will update the backup copy.
Metadata changes (such as adding keywords, titles, etc.) will update some file types — most commonly JPEG, PSD, TIFF and DNG. We can’t avoid using the first three, but that’s why I don’t recommend using DNG to store your raw files (stick with the proprietary format for now — metadata for them is stored using small .XMP sidecar files).
I know better, but I’ll still forget and make significant keyword changes to huge numbers (many thousands) of images in Lightroom. Most of those are written to the small .XMP files, but there are still lots of big TIF/PSD files in my catalog, so when I do this I immediately generate a lot of changes to feed to the backup service.
Considerations for an online/cloud backup service
Platform support and price are obviously factors between the services, but we are all faced with the same constraint — basically we’re talking about time (constrained by upload bandwidth)
Any system running your backup needs to run long enough to let the backup complete — that’s going to be a challenge if it’s your laptop and you spend 1⁄2 your day disconnected and then leave it powered off when you get back to base.
If your daily data growth (either new images or changes to existing data) exceeds the data that you can upload before you start adding more the next day then you’re stuffed.
Your backup is also running in parallel with everything else you do, so if you need to publish files to your website or upload an album to your print lab that’s going to be slower (and your backup will slow down too).
Making a cloud backup solution work
To make an online backup work you’ve got to have decent upload bandwidth (relative to your original data volume and daily updates).
In normal use you have to be able to leave the backup client system running 24x7. This is where I have the luxury of dedicated backup host with redundant/resilient storage — I copy my local working storage (a portable 6TB USB3/ Thunderbolt 2 brick) to the backup server, leaving it running and allowing me to switch off my desktop machine and disconnect my primary local storage.