The self-described "systemd cabal" outlines their long term plans for re-architecting the design of Linux systems (0pointer.net), primarily using btrfs features, along with systemd. While I've personally been largely skeptical of systemd's implementation, I have to admit that this vision sounds very nice, and extremely doable with some buy-in from larger distributions. I'm sure there are a lot of drawbacks also, but as long as some of the zillion distros stay with an old-school approach (and I'm sure they would), I see nothing wrong with having the choice of a purely modern architecture as they describe.
Some key points are:
utilizing btrfs subvolumes and snapshots to compartmentalize applications for ease of installation (as well as development (devloping against standardized libraries), rollback, security.
user home directories as btrfs subvolumes and having system users (think /etc/password) extrapolated from the information in the namespace of those subvolumes.
Easy mix and match of operating systems, os levels, application versions with rollback.
Systemd Master Plan Outlined by the "systemd Cabal"
from the not-the-traditional-linux-way dept.
CoolHand writes:
The self-described "systemd cabal" outlines their long term plans for re-architecting the design of Linux systems (0pointer.net), primarily using btrfs features, along with systemd. While I've personally been largely skeptical of systemd's implementation, I have to admit that this vision sounds very nice, and extremely doable with some buy-in from larger distributions. I'm sure there are a lot of drawbacks also, but as long as some of the zillion distros stay with an old-school approach (and I'm sure they would), I see nothing wrong with having the choice of a purely modern architecture as they describe.
Some key points are:
Post Comment