[OpenIndiana-discuss] What is OI/illumos stance on recreation of features from Solaris 11 (and other proprietary OSes)?

Jim Klimov jimklimov at cos.ru
Wed Oct 10 08:51:37 UTC 2012

Hello all,

   I was at an Oracle presentation regarding Solaris 11, and they
did mention a number of features which I think appeared after the
code split. Some sound like ideas simple enough to be recreated
without using their source and just make a look-alike, with any
other conincidence being purely coincidental.

   The main question is IMHO legalese FUD: if we go this path and
recreate in illumos from scratch look-alikes of nifty features
and minor improvements that were created elsewhere, are such
works vulnerable to lawsuits?

   It was interesting to learn also that ex-Sunnites at Oracle
are (sad to be) forbidden to even look at alternate post-split
OSes so as not to become "tainted" by the knowledge, and that
any import of the code is also forbidden - if needed, things
can and must be recreated in-house as original works. The main
concern is post-SCO debacle: if someone posts non-original
work into illumos claiming it's their own, and then Oracle takes
it as CDDL code, and then the original/new IP owners sue the
world... this is what Oracle wants to avoid. Not that they are
inherently evil and arrogant (well, not everyone - to be sure).

   So, the main question of this post is: in legislations that
are relevant and actively applicable to illumos-derived OSes
such as OpenIndiana, are *ideas* and technical specs (meaning
public-API/config compatible but code-different implementations)
patentable and protectable, or are coders allowed to implement
on their own things that others created somehow else?

   Among the features that looked interesting and might be missing
in the open-source releases are:

1) ZFS dataset encryption, as is long discussed and longed for
    by community members and users;

2) Shadow migration (and/or hierarchical-storage-management sort
    of overlays) that allows userdata to be reorganized between
    datasets and whole servers transparently to users, who can
    bang their read-write IOs against the storage server farm
    or local disks during the migration process.

3) Immutable zones (several levels of immutability, with part
    of the zone filesystem being read-only) - basically we had
    much of this implicitly with sparse-root zones, but the new
    feature is a configurable attribute of the (fullroot) zone
    and allows to permit writes to /var+/etc, to only logs, or
    to nothing at all, basically.

4) Automatic VNICs for zones - if I got it right, if a VNIC
    if referenced in config of an exclusive-IP local zone and is
    missing in global zone Crossbow config, the VNIC is spawned
    temporarily during zone boot and torn down at its shutdown.

5) An enhancement to "zoneadm" that allows to gracefully shutdown
    a local zone from GZ (essentially "zoneadm shutdown {-r}" seems
    like just a shortcut to "zlogin zonename 'init 5{6}'").

   Of course, there are hundreds more of features that were not
even touched in the presentation, but of those that were - some
of the above look scriptable and simple; others less so, but
still quite useful.

   MAY they be implemented for illumos-gate/OI?

Thanks for ideas,
//Jim Klimov

More information about the OpenIndiana-discuss mailing list