nydus
nydus copied to clipboard
Nydus sub-blob garbage collection support
In order to support container OS upgrade, each increased blob stores the data difference of previous container OS images.
However, the number of such blobs can finally be very huge which impacts performance and even cannot work properly anymore, so we need to decrease the number of total blobs.
Why there is an impact on performance if layers is getting increased.?
Why there is an impact on performance if layers is getting increased.?
since we need to mount (or open all blobs) in advance. The description is still incomplete for now(just used to share a link to who is needed), I will fill the full request later.