[NTLUG:Discuss] Re: Practice with clustering at home? -- summary of external busses ...
Bryan J. Smith
b.j.smith at ieee.org
Tue Aug 30 16:38:06 CDT 2005
Thomas Cameron <thomas.cameron at camerontech.com> wrote:
> I appreciate your response, but it has nothing - absolutely
> nothing - to do with my initial request.
> I understand the differences between storage technologies.
Your initial USB comment suggested otherwise to myself --
that you had never used FireWire. From there, maybe I
assumed too much -- but FireWire has _never_ had the ability
to be "multi-host targettable." People confuse its
"multi-device targetting" for "multi-host targetting" -- it's
not the same.
> What you fail to understand (or refuse to acknowledge) is
> that what I am doing is purely academic and requires a
> shoe-string budget.
I'm just trying to let you know
> I'm asking to crack a 3-minute egg and your reply is to use
> a sledge hammer.
No, quite the opposite.
You're using the sledge hammer because it's cheap. I
recommend you look to a neighbor for a custom tool that he no
longer needs that would work much better, and you could get
for not much more.
Plus SAS is really going to make things rather commodity
really fast. It's finally what the storage world has been
asking for a long time, and is finally getting.
> It is really coming across as a know-it-all, "look
> at how smart I am" response.
I think this is a good example of why I don't belong in this
group. Because my recommendations are taken that way,
instead of just being taken as what they are ...
recommendations. They are not made for reasons of arrogance
or whatever assumptions select people want to make.
I'm sorry I'm verbose. I write for part of my living and
that comes with the territory. I try to curb it, but I find
the complaints are just as bad.
Because in the end, all I'm trying to encourage you to look
to some other, low-cost means. A lot of times people assume
it is expensive to do it "the right way" -- when there's a
lot of used hardware out there.
The new, popular kernel hack that attempts to use a FireWire
device by multiple hosts _ignores_ the fact that the
end-device is still assuming it's exclusively used. That's
the core problem, not the bus interface or how the hosts work
-- the end device.
So what you learn from this will not only be incomplete, but
rather inapplicable. All the setups are good for are little
more than just getting an environment where you can develop
(e.g., an Oracle developer) -- not actually learn how to
configure a real-world setup (which you'll have to learn
differently).
--
Bryan J. Smith | Sent from Yahoo Mail
mailto:b.j.smith at ieee.org | (please excuse any
http://thebs413.blogspot.com/ | missing headers)
More information about the Discuss
mailing list