[lug] RAID installation on Fedora 6 Zod
Nate Duehr
nate at natetech.com
Thu May 17 17:15:24 MDT 2007
> But back to the original points. Yes, Linux software raid is maybe a
> 90% solution, but it's still a decent solution for a lot of folks who
> can't afford the 100% commercial solution.
But that *was* my original point -- that if you can afford it, avoid Linux
software RAID and save yourself the headache of upgrading again later --
if you have the budget and understand the real needs of your business...
if the goal is 100% uptime, Linux software RAID is a stepping stone you'll
eventually pass over on the way to something better.
If the goal is to mirror a home Linux machine, and the 90% solution works
for you, great... but the original poster was talking about building a
server. If the money is spent today, the system will have the 100%
solution, from day one. If the 90% solution goes in, and you really
needed the 100% solution -- you'll be spending more time and money
re-doing it later.
So I never disagreed with what you're saying -- I was simply offering to
the original poster that there's always the option to do it right to begin
with -- if that's what his real needs/desires are for his business.
It's a similar sentiment to a real-life problem... If I had thought about
my required uses for vehicles more carefully, I would have saved a lot of
money buying a diesel pickup truck in 2000 vs. the Jeep I "settled for" to
ease the up-front costs. Now, I'm being frustrated that the money to
"upgrade" doesn't make any sense right now... I'd be 125,000 miles into a
vehicle that would easily last 300,000 instead of 125,000 miles into a
vehicle that will become more and more questionable as to how much I can
trust it in the back-country. I would have spent more up-front, but
wouldn't be shopping again in two to three years. Cost, higher on the
pickup option, no doubt. Headache/hassle, lower... because I'm constantly
borrowing large haul vehicles and trailers over the lifespan of the Jeep.
It's not as easy to borrow a friend's server when you land that big
customer that forces you to sign an SLA. So in the end, you end up paying
now... or paying later... might as well buy top of the line technology vs.
twiddling with something that's "close" and then having to buy top of the
line later, anyway.
Does that help make my original position make more sense?
--
Nate Duehr, nate at natetech.com
More information about the LUG
mailing list