raymor

I see you are new, only been registered about a year, so I'll explain
to you some of the reasons that all of these pros who have been
doing it for ten years or more do it the professional way. They are
some of the same types of reasons that professional chefs don't
build their own oven out of a hair dryer and an ammo box, though
that seems like it would kind of work, but I'll give you some specifics anyway.

Plugging in a "server" which performs like crap and gets hacked within a week
is indeed a no-brainer. On average, such a box will take about 2.3 days to get hacked.
(Based on honeypot experiments). So if you're wondering where all that spam is
coming from, it's coming from your "server". Web hosts pay their top admins $100 / hour
and up because doing it right is far from a no brainer. I have a 456 page book here
just on the topic of diagnosing crashes on servers. I have two more books totaling
over 1,000 pages on the Apache web server, and probably several thousand pages
covering MySQL. Those books don't contain half of what I need to know when a
host calls me with a problem their guy can't figure out quickly enough when members
are canceling in droves because the site is down. So it's a no brainer only if you
define no brainer to mean something that only requires 5,000 pages of reading to
be good at. You could learn all that stuff - or you could learn how to build and market
sites to actually make some money. Professional web site owners generally find that
it makes sense for them to spend their time and energy making money from their sites
and pay Phatservers a few bucks to know exactly how the Timeout directive interacts
with the MaxClient directive to avoid DOS. What happens that way is you get to share
a couple of top notch admins with several other webmasters. Mike, Chris, or Matt at
Phatservers are $100-$200 per hour type admins, so you get experts running your
server, but you pay almost nothing for it because there are a few hundred other
webmasters pitching in to pay Chris's paycheck with you.

Similarly too for DSL providers and any other home connection - they will eventually
catch you if you get any significant traffic. Then not only will your site be down, but
you'll have no home internet connection to use to find an actual web host, upload your
site, etc.

Of course DSL over either phone line or a cable line is only going to have enough
upstream bandwidth to handle one or two customers on your site at once and keep
them happy. Remember they too have a cable connection and they expect your site
to load at 1Mbps at the very least, preferably more like 4Mbps. If you have several
people on the site at once your cable or DSL isn't going to cut it. If you do NOT have
several visitors on your site at once, you REALLY need to take some time to design
a solid site and do some marketing to get traffic to it. That'll be a LOT more profitable
for you than spending your time figuring out why you server can only handle a few
people at once. (Like because you haven't set MaxClients right and are loading 20
modules which you aren't using).

Go check out a real datacenter some time and you'll see a few other reasons.
At home, your power probably goes out for a little while once or twice a year,
and your internet connection another once or twice when a tree limb falls on a
phone line or something. Many of my neighbors were without internet access
for nearly two weeks recently after a storm. Just last week another friend,
and much of Texas and Lousiana, was without internet for three days when a
fiber line was accidentally cut. That's OK for surfing, but for a
professional site down time means customers who would have rebilled another
six months cancel.
At a typical datacenter like one we may colo some servers at, primary power
comes from two different power grids. If both power grids fail, there are two
racks of 100 batteries each which power the DC for a few minutes until
the generators take over. The power outage may have been caused by a
hurricane, tornado, or flood, which has torn up the surrounding area and those
generators will eventually need more diesel fuel, so they have that well covered.
The diesel truck can pull up next to the building and pump in diesel. Failing
that, there's another diesel connection on the other side of the building to
pull diesel from a train. Failing that, a third diesel hook up is on the roof, next
to the helipad. This is the arrangement for power at the datacenter we're
considering. By the way, notice a said generatorS plural - three generators
each capable of powering the facility alone should the other two fail.

Similarly what happens if someone breaks into your house while you're
at Wal-Mart? Your business is gone, just like that. At the datacenter, they
use a Medeco lock on the first door and retina scanner on the second -
you're not getting to my servers unless you get my eyeballs first. Why
two doors, the first and the second? For the two buildings, an ultra modern
concrete building built on a bed of springs inside of an older building.
That way if a tornado damages it's OK, the servers are inside the concrete
building which is inside the building which was damaged. The raised
bed of springs protects from flood, earthquake, and shock from explosions -
either the nearby factory blowing up or even a nuclear device downtown.

All of this costs just a little more than the extra power and air conditioning
bill would at home, and my server is only down when I take it down.
So hosting at home will work, sort of, until either your site starts getting
traffic, or there is a good storm, or someone breaks in, or a dozen other
things. As soon as something bad happens, you'll be out of business and
understand why professionals do it the professional way, working with
other professionals.

__________________
For historical display only. This information is not current:
support@bettercgi.com ICQ 7208627
Strongbox - The next generation in site security
Throttlebox - The next generation in bandwidth control
Clonebox - Backup and disaster recovery on steroids