Barry Roberts wrote:
If Coraid gets sued and goes under, who's going to repair failed
Coraid hardware? Sure, the protocol is open, and the drivers are open
source, and the drives are industry standard. But that doesn't make a
bit of difference if you've got a few thousand AoE controllers, no
remaining manufacturer, and they start failing.
Single source hardware is risky no matter how much they wave the "Open
Standards" flag. If AoE catches on and there are multiple, competing
sources of the hardware, I have several applications where I would
love to use it. Until then, it's nowhere near open enough for me, and
I'm not a PHB.
Let's say you have a lot of Coraid hardware and Coraid vanishes. If you
already have a NAS head and you have to add another, you take Coraid's
software and build a box that does exactly the same thing as a Coraid
NAS head. That's probably not difficult.
If you have to add a shelf, you add a shelf of any variety that can talk
to Linux, then you set up a Linux box between the shelf and the NAS
head. I assume this is possible. If the current kernel doesn't provide
any way to translate AoE to other block protocols, at least all the
specs are available. (Also note that performance might not be
top-notch, but if speed is really the main factor then you shouldn't be
using gigabit ethernet for disk access anyway.)
If some of that turned out to be a lot of work, you create a new little
company to serve the needs of former Coraid customers. The barrier to
entry is low because of open standards.
Shane
/*
PLUG: http://plug.org, #utah on irc.freenode.net
Unsubscribe: http://plug.org/mailman/options/plug
Don't fear the penguin.
*/
