I would like to use kopia to back up my second hard drive on a regular basis. This has the following structure/directories:
d:\data
d:\video
d:\images
d:\projects
d:\archive
I want to backup everything except “d:\archive”.
Do I create a snapshot for the complete disk “d:\” and exclude the directory “d:\archive” or do I create a separate snapshot for each directory “d:\data”, “d:\video”, “d:\images” and “d:\projects” ?
What are the advantages/disadvantages of the two options?
And regarding compression:
Why is by default no compression enabled?
Is there a recommended setting for compression if i want to use it?
Using multiple snapshots allows you to apply different policies (retention, compression etc.) but might become complex if you add more snapshots later on.
Probably for performance reasons and because it doesn’t always make sense to use compression. E.g. I do have repositories where just 10-15% of the data is compressable.
Is there a recommended setting for compression if i want to use it?
And is something to consider, when i work with Windows and Linux (Dual-Boot on Disk1) and want to make backups - sometimes from Windows and sometimes from Linux? (Both times the same directories from the second hard drive to the same repo.)
There is an article about compression in Kopia in the documentation:
I mostly use zstd-fastest as it provides good compression at high speeds and low memory usage. But you should do a kopia benchmark compression yourself and pick whatever suits you.
The settings are very much varying depending on the kind of input data you usually have, but also they cater for very different priorities, either
“compressed size”, or “compression speed” or “ram usage while compressing”.
Also, different CPUs (both type and number of cores) will affect how well they do different kinds of compression so one size will never fit all.
Actually, it is the same for splitter, crypto and hash algorithms, but they must be the same within the repo, so once they are set, they can’t be changed so that is why those are not a per-client choice.
Yes, a 64-core machine can ask for its data to be compressed using zstd-best, and a slow single-core 100MHz-cpu box could ask for its data to be “s2-default” or some other compressor that doesn’t pack very well, but is lots faster.
Of course they will not deduplicate against eachothers data, but they can use a compressor that fits the needs of that particular computer.
(and my slowest kopia-using box is not really 100 MHz, but dual-core 500MHz octeons)