[lug] Hosting Question

[Sean Reifschneider]

On Fri, Sep 29, 2006 at 03:53:58PM -0600, George Sexton wrote:
>issue of fault tolerance. I keep spare drives and power supplies on hand 
>for my servers. But, if a server dies then I'm on the hook for getting a 
>new one up to the facility.

This is one of the benefits of a "managed dedicated" arrangement over
co-location.  Back in the early days of our hosting, we dabbled with
co-location, but we weren't really set up for un-escorted third-party
access, and the colo boxes tended to have far more hardware problems than
our own machines/  Plus, when there was a problem, since we didn't own the
hardware, we couldn't really do anything about the problem.

We decided instead to concentrate on the managed hosting, which at that
time had very little competition.  We have spare hardware available, so not
only does our staff get alerted if there's a problem at 3am, we have a full
machine we can drop the discs into to get things back up and working right
away.  We have the option to replace the whole machine, instead of spending
time while services our down trying to isolate the problem.

This is one of the benefits we get from scale.  Unless you're co-locating a
lot of boxes, most people don't drop in extra hardware (taking up extra
space).

Managed hosting is definitely something I'd recommend people look into.
These days there are a lot of choices.  From inexpensive covering only
hardware, to $1,000 per month or more covering all sorts of software
maintenance and management.

>The biggest problem I had with self-hosting is power. It seems that 

Yeah, I hear you.  Our house is on "rural electric" and in a year we'll
probably have one or two dozen power outages, usually in the few second
range, one or two in the 20+ minute range.

In comparison, our hosting facility pulls power from two different Denver
city sub-stations, and has diesel generators with 7 days of on-site fuel.
We've designed all of our machines to run off redundant power feeds up to
the in-rack PDU, so up to 4 feet from the machine there's no single point
of failure.  If that's not good enough, we can offer you a High
Availability cluster with no single point of failure.  :-)

Power is one of those things that it's nice to have solved.

Sean
-- 
 Put out fires during the daytime.  Do your real work at night.
 Sleep is just an addiction.  -- Dieter Muller
Sean Reifschneider, Member of Technical Staff <jafo at tummy.com>
tummy.com, ltd. - Linux Consulting since 1995: Ask me about High Availability