[Hcoop-discuss] Next steps for HCoop

[ntk at hcoop.net]

I'd like to make some notes on this whole thread.

First, I do think it would be both ridiculous and not very reliable to
submit a Slashdot question in hopes of answering it with an ad for HCoop. 
I also think people are off base if they think we will suddenly get
thousands of serious applications on the basis of a single +5 comment.

Having more people applying than we can accomodate is not actually a
problem.  In the worst case we can always do what we did before and
waitlist people if we max out our capacity.  However adding we can add new
servers by more or less cloning fyodor more easily than we set it up to
begin with.  The #1 issue I see at this point is changing the portal
software to be aware of multiple servers and all our users, we don't want
to have separate portals for each box and we want to keep track of user
resource usage across servers.

I'd also like to throw out the idea of looking beyond the typical "Linux
geek" type of audience (of which I and most of our current users probably
qualify as members) for new users.  Certainly we are currently a good fit
for such users, but I do not agree with the sentiment that they are the
only ones we should target or even the only ones that we are currently
useful to.  More on this in a moment.

Also a couple of people have proffered that we can't compete on cost with
the lowest-cost providers out there and instead should emphasize
flexibility and other advantages.  Actually, looking at all of the
services linked to so far, all of the plans that are actually cheaper than
our base dues have severe limitations compared to HCoop (in terms of
numbenr of domains, disk space, applications, etc.)  I submit that due to
not taking any profit we are already cheaper or as cheap as any other
commercial provider providing the same level of service (in terms of
unlimited domains, amount of disk space, cost of bandwidth, shell access,
CGI & other web apps, databases, etc.)  Moreover, as we add MORE users we
will only continue to get LOWER cost.  Adam already estimated that we can
accomodate up to 300 users at current usage levels on fyodor.  This would
at most only moderately increase our marginal costs (bandwidth, basically)
while reducing our dues by a factor of 6.  Moreover, once we DO get to the
point of leasing a whole block of units or a whole rack, our monthly costs
per U will be significantly lower, providing yet another level of savings.

So, even users who do not take advantage of all the "geeky" things that
hcoop has to offer benefit from HCoop and provide a benefit.  By
increasing our volume they increase our efficiency, providing savings to
them and us.  There are plenty of people out there with basic computer
skills who, once they learned to synchronize their web directory and check
email (either plugging a few settings into Outlook or better, Thunderbird,
or simply using squirrelmail), would be satisfied customers.  In fact, I
contend that such "neophytes" are actually LOWER maintenance than
potential "UNIX wizards" who come along and want to do amazing dynamic
things but end up needing a lot of assistance getting their Postscript web
server running and proxying their backend cluster running Plan9 in
Kazakhstan.  A little knowledge can be a dangerous thing.  Either way, the
"neophytes" certainly use fewer server resources than the power users,
which means either lower costs for everyone, or an opportunity to provide
an even cheaper low-cost alternate plan for such users.

-Nathan