[Discuss] SysVinit vs. systemd

[Bill Ricker]

Can we agree that SysV Init is ancient, and is due for at least major
reform, if not outright replacement ?

There have been several boot acceleration / init rationalization
attempts recently.

Some have been init reform. Efforts to reform Init with automagic
learning for parallelizing of boot order seemed promising to me, but
have fallen by the wayside. I'm unclear why - did they fail or were
they successful but even in aggregate insufficient reform? Or did the
Big Distros suffer NIH ?

Other efforts have been aimed at near-greenfield replacement. Upstart
was Canonical's ironically-named independent effort, for which they
took quite a bit of flak (it smelled like another MIR NIH
ourwayorthehighway project), but they've belatedly hopped on board
with systemd now, after Debian voted (by tie-break) for systemd. [1]
   * SystemD is at least supposedly backwards compatible with SysV and
LSB init files?
  * Has Debian indicated what they intend to do for their BSD and Herd
kernel builds, since SystemD requires linux-specific kernel features?

If one wants to review the technical pros and cons, the Debian Init
debate is probably the best place to look. [2]

>From the paranoid security-is-the-only-requirement point of view, one
daemon to-rule-them-all breaches compartmentalization. Speed at the
cost of security and reliability is a bad bargain. Systemd's use of
Cgroups should help here, if properly done? (And the RedHat
SELinux/Docker team will keep an eye on that.)

While we can debate artistic design choices and complain about
personalities, it seems we don't have a lot of choice: if RedHat and
Ubuntu and Gentoo and Debian will all be shipping SystemD, only the
most fringe distros will keep SysV init on pid 1 or find a third path.

-- 
Bill Ricker
bill.n1vux at gmail.com
https://www.linkedin.com/in/n1vux

reference:
[1] Mark Shuttleworth » Blog Archive » Losing graciously
 http://www.markshuttleworth.com/archives/1316
[2] https://wiki.debian.org/Debate/initsystem