Vidar, > RAID is a buzzword for enthusiasts. A lot of people think RAID will give > them better gaming performance and faster downloads. Windows doesn't do > software RAID too well, and a fullblown hardware RAID is too expensive. > Creating these monsters was the answer. > (my own text deleted) > Hotswapping depends on the controller, AFAIK. Very few SATA controllers > out there support hotswapping, especially the cheap ones. > > I researched the subject of fakeraid a few years ago for a workstation > with a cheap "RAID" controller. On the kernel mailinglist I read Linus > saying something along the lines of (this was about Promise controllers) > "we will not implement support for fakeraid on Promise controllers. > Software raid is better done by kernel than driver and Promise agrees to > that". (completely from memory, not an exact quote!) > > The point being that the *drivers* do the RAID which is a mess when you > want to administer the RAID. Instead you let the kernel do the handiwork > instead of making bloated drivers will full RAID functionality and no > tools (or one tool per driver... *shudder*) to administer them...
Thanks for all this development. You've convinced me, I'll probably give up on the "firmware RAID" drivers. > > Since our standard is TSL 2.2 I'm heading towards a customized > > installation of 2.2, possibly with a 2.6 kernel and yes, Linux > > software RAID. > > This sounds far worse than loading TSL3, but you probably know what you > are doing. Well, we have a server farm and the standard O/S now is TSL 2.2. Don't want to move away from this until TSL 3.0 is called the stable, production release by the Trustix folks, which it wasn't last time I've checked. Also when we have a general migration plan. We already have a few boxes running a 2.6 kernel on top of TSL 2.2 and we've been quite happy with this setup so far. Thanks again, Greets, _Alain_ _______________________________________________ tsl-discuss mailing list [email protected] http://lists.trustix.org/mailman/listinfo/tsl-discuss
