Re: Solaris alltogether needs to die
Back in the day when I entered the corporate IT world, Solaris 10 was at its peak. ZFS was a distant ship on the horizon, disk slicing was necessary -seasoned with some Veritas to make the life of a SysAdmin bearable- and the separation of the swap space and /tmp wasn't a requirement of the design documents put together by The Powers That Be. Mount some development zones on top with mediocre coders messing with them and you have a recipe for disaster.
Not to mention the comparative cost of operation for a comparable POWER frame was lower and you had effective VM separation, instead of those early attempts of containerization called "zones".
I'm not denying Solaris was a foundational piece in some of the technologies we enjoy today, but some of its core requirements (maintaining backwards compatibility with obsolete prior versions to name one) made it harder to administer than its rivals. Even more with some incompetent head architects. Let's face it: playing the old, tired card of "back in the day you really had to know your ways around $PIECE_OF_SOFTWARE" only speaks about your grumpyness.
And I'm not talking about dealing with a couple servers, of which you could eventually get to know every single quirk. I'm talking about ruling over three hundred physical servers, with LDOMs and zones so poorly distributed (dev/staging/prod on the same blade) you'd learn to hate Solaris the way I did.
Again, when we finally got off them and flew to AIX 6.1 and Red Hat with XFS support, we couldn't believe how much our SysSdmin lives changed.
Icon because I think a chat like this deserves to be enjoyed along a good one.
Biting the hand that feeds IT
