[Hcoop-discuss] Help plan our next server set-up!

[ntk at hcoop.net]

> So bear with me...what benefits am I getting at the moment from keeping
> Abu going online?

I think others are more qualified to answer that than I am--I'm not
officially an "admin" nor do I even have root passwords on either abu or
fyodor.  If your question is what direct, immediate services are you
receiving from Abulafia at this moment, I believe the answer is "none,"
unless you count backup DNS, which is only of much practical use if you
are using hcoop to serve DNS for sites not hosted on hcoop.

But before you conclude that we should just take it down right away, it's
a little more complicated than that.  We could put more services on it,
perhaps backup MX, which would be very good to prevent bounced mail during
network or other outages.  And as Shaun pointed out, right now if fyodor
were to go down altogether, it would be minimally several days before we
rebuilt it from scratch (and hopefully you've backup up your files &
mail!).  Abu isn't really capable of being a hot-swappable replacement for
fyodor, but it could be used as some sort of temporary replacement.  I'm
just really just brainstorming here, but it does have potential future
value.

Moreover, if we take it offline, we would have to pay to have it shipped
to us.  If we ever wanted to put it back online either with Xiolink or
elsewhere, that would involve shipping it again and probably paying
another activation fee.  All told those expenses would probably be more
than we could make off of selling it.  So unless and until we decide that
Abu is no use and we are taking it down permanently, it probably costs us
less longterm to pay for hosting it now.

But the #1 thing it is useful for is just a generic Linux shell server for
development and whatnot.  I know Adam said after the migration that
everything should be moved over and he might close login off (which was
about the only thing that got users to move over to fyodor), but as far as
I know everyone who had a Abu account can still log in there.  I have
occasionally used it to compile code and run small computations, test
scripts, and also running an IRC client to access #hcoop when I'm on a
machine that doesn't have one--which is handy since Interserver blocks IRC
both incoming and outgoing.  I think it might be worthwhile to at least
temporarily christen Abu as a "development box" and have it as an actual
policy/recommendation that users go to Abu for things like testing,
compilation, anything that would tend to burn memory & CPU-type resources
but not lots of bandwidth.

But regardless, we have no choice but to split the costs among hcoop users
if we are to keep it online, which leads to...

> Now if keeping it going meant that one would (optionally?) pay $1.50+-
> month extra for some specific features (that I might not need or ever use,
> but others depend on), or level of service (backup in case of primary
> outage), then I'd agree...but right now, for my limited needs (a local
> community chat-room), I don't see the advantage.  So $1.50 isn't much,
> true, but for me, it would get me 25min of voice calls to users (most
> important) or 30 pages of copies or 3 stamps.

The problem is that HCOOP has to pay the whole sum for hosting Abu every
month no matter what.  It's not like it costs us $1.50 per user to host
Abu, it costs us a flat fee of $50+bandwidth to host it every month.  So,
if we made it "optional" and only five people "opted" in, then they would
each be paying $10 a month.  But if we ended up using Abu for a purpose
that benefits all users, then those who opted in would have shouldered a
grossly disproportionate share of costs.

We can't just carve up HCOOP's overhead expenses piecemeal and ask members
to pick and choose which they want to pay for.  I think once we're bigger
and have more hardware, it will be easier to provide a range of services,
perhaps from basic shell+X gigs bandwidth web hosting for $5/month up to
linux virtual servers for whatever extra it costs us to provide that
service.  But for now, we have to divide the expenses we have.  The only
real difference between one user and another right now is bandwidth and
disk usage, both of which we are underutilizing at the moment.

-ntk