|
|
| Author |
Message |
steve
|
|
The Growth of Grex
| 
Jan 7 04:06 UTC 1996 |
I was curious where the new users to Grex were coming in from,
so I did some digging into the newuser logs, and made the attempt
to figure this out, based on the input from the last 1000 runs of
newuser. It can't be 100% accurate, because of the way that people
enter things in. Since some people didn't specify anything useful
at all, I didn't count them at all. But it looks like 714 out of
the 1000 gave useful enough information to be included, or better
than 2/3's, which is about what it was a couple of years ago when
I did this last.
The results really kicked me in the head.
I think the information here says some things that we need to
address. Overall, we're doing a fine job of publicity. We're
still getting 50 - 70 new users a day, and the passwd file now
has just under 9000 entries. Lately, I've been killing off
slightly fewer accounts, which means that more people are finding
Grex of enough value to hang around. We do need to make more of
an effort at getting locals in, since they've become about as rare
as users from England.
Unfortunately, with 95 members and 8,906 human accounts on Grex
this means that the percentage of paying members is 1.06%. Put
another way, 1 in 94 accounts here is a member. Now, remember that
some of these accounts are the 'use once and forget' type, but
still, if we consider *half* of all the accounts here of that
variety, then we're only at 2% of the users being actual members.
We're hovering at around 100 members right now. If we could get
to the 500 member mark, we could afford to pay market rates on a
T1 connection for Grex, pay the bills and still have something
left over for expansion.
I think we have the possibility to do this, because the raw
numbers say to me that we have one hell of a lot of people
wandering in here. Lately we've been having disk problems, and
we still don't have usenet available, but usage is still going up.
Add to this, the fact that the rest of the world is discovering
Grex. In the table below, look at the top 10 entries:
Note where India is. Note carefully. To me this is both
exciting, and really scary. What is Grex going to do when the
world discovers Grex? I pretty much mean that literally now.
India recently allowed privitized Internet connections to exist,
and the market there is expanding *rapidly*. Their prices aren't
even that bad: about $US 2.50/hr for little accounts, and there
hasn't been real competition yet. I spoke to two people recently,
one from Thailand, and one from China (China?!?), and these people
said that the Internet was about to hit in their countries in a
big way soon. Probably Thailand first, since the Chinese government
still wants to put leashes on everything, but they know they can't
hold out much longer. The recent international women's conference
held there created too much exposure to the Internet, and now that
many thousands of people there have seen it, it won't be that long
in coming. I personally predict less than 2 years.
So what do we do when the world gets on Grex?
I pose this as a real and serious question. I used to joke
about things like all the uid's being used up because everybody
wanted on, but now I don't think it's so funny. Neat maybe. And
definitely possible. It used to be possible to do systems type
things in the morning usually, but no longer. In the dead of
night when most American users are asleep and Eupopean users are
working, Asian users are streaming in. There used to be a time
when we could take the system down and not impact that many
people. No longer.
We're going to have to start thinking of Grex in a different
way, I think. Up to the point of being on the net, we knew we
were tiny, and that was that. There simply wern't that many
people who'd call us long distance, so we knew that we'd hit
some form of saturation, and that would be about it.
Not any more. The number of people who will be on the net
within three years is nothing short of amazing--quite likely
there will be 1 billion people in cyberspace. If only one in
two thousand discovers us, that means we have a potential of
500,000 people passing through Grex.
I think we need to seriously think along these terms, because
we're seeing the the beginning of it already.
So, I'm thinking of the following things:
1] How might Grex have to adjust itself with a user base
of 100,000 people? Can we stay as open as we are now?
2] What is the best way to get the membership that we'll
need to get to something like a T1 connection?
3] Do we *want* to do this? What will we lose if we grow
by another factor of 10? What do we gain?
--------------------------
Country approximate number of logins from that country
USA : 274 (except Michigan)
India : 128
Michigan : 49 (anywhere other than the 'local call' area)
Canada : 33
Local call : 26 (Ann arbor, Ypsilanti, Saline, Chelsea, etc)
England : 21
Austrailia : 15
Malaysia : 14
Singapore : 14
Italy : 12
Mexico : 9
Netherlands : 9
Thailand : 8
Japan : 7
Germany : 7
Spain : 6
Turkiye : 6
Brazil : 5
Ireland : 5
Israel : 5
Poland : 5
Sweden : 5
Indonesia : 4
New Zealand : 4
Norway : 4
Russia : 4
Bulgaria : 2
Chile : 2
China : 2
Columbia : 2
Country312 : 2 (312 being the country code, but I don't know what that is)
Croatia : 2
Czech Repub : 2
Finland : 2
Phillipines : 2
Austria : 1
Bahamas : 1
Barbados : 1
Belgium : 1
Bolivia : 1
Denmark : 1
France : 1
Hong Kong : 1
Iceland : 1
Korea : 1
Latvia : 1
Malta : 1
Pakistan : 1
Puerto Rico : 1
Taiwan : 1
Uruguay : 1
Venezuela : 1
|
| 239 responses total. |
kaplan
|
|
response 1 of 239:
|
Jan 7 04:33 UTC 1996 |
Wow! Of the Michigan and USA catagories, can you tell how many of them
are in or near the "traditional" grex population centers like Columbus,
McKendree, and Gaylord?
What was the date that people from out on the net were permitted to telnet
in? How many of those 8000+ humans in /etc/passwd have been there longer
than that?
|
rcurl
|
|
response 2 of 239:
|
Jan 7 08:10 UTC 1996 |
Do your data tell what these users are doing here? I do not think there
is a very much larger group of users participating in conferences than
there were a year or more ago. That means the influx of newusers are
making other uses of Grex - perhaps party, perhaps just e-mail, perhaps
to have a free web page?
|
robh
|
|
response 3 of 239:
|
Jan 7 12:08 UTC 1996 |
Most likely a little of all three of those.
|
carson
|
|
response 4 of 239:
|
Jan 7 12:48 UTC 1996 |
I think both Gaylord and McKendree have fallen off a bit as far as
population centers, but that Columbus is still going strong. The
internet connection was opened sometime shortly after I created
my account... I've got nothing else to do, since there aren't many
people on, so I think I'll do the math by hand. :)
|
carson
|
|
response 5 of 239:
|
Jan 7 12:57 UTC 1996 |
about 95, if moo is our earliest newuser from the Internet. He's just
a couple of place below me. I think the most interesting thing I learned
from that glance was that kaplan created his account after me. ;)
Lessee... 95 out of 9000... about 1%. Strangely enough, that's also the
number of members. ;) wow.
|
gregc
|
|
response 6 of 239:
|
Jan 7 14:13 UTC 1996 |
Frankly, the idea of getting a T1 for Grex scares me:
1.) We don't have enough computing power to handle a T1.
2.) Even if we got enough members to pay for a T1 *and* enough new CPU
power to handle it, we simply do not have the *manpower* to manage a
system of that scale.
3.) Manpower is the problem here. With more members we can buy bandwidth, we
can buy CPU, but are we going to buy(ie: hire) manpower? By comparison,
the cost of a T1 and a big CPU is *cheap*, when you consider that hiring
1 or 2 full time support people could cost over $50,000/year.
4.) Do we *want* to get that big? I think not. One of the reasons that Grex
can be run in the casual, freindly way it is now, is becuase of it's
size and it's *lack* of any significant amount of money. Turn Grex into
a big system, with hired people and an annual budget of $100,000+ and
you'll see things that make the current Mnet look *tame*.
Yes, we need to get a faster link, but I don't think we want to go beyond
a 64K or 128K link. Otherwise we'll end up with something that is completely
unmanagable with our current resources. To cope, we would have to change
Grex into something that I think most people, staff/members/users, wouldn't
like.
The solution is, that at some point, we are going to have to set a limit
on the number of user accounts. Even if that limit is 40,000, there will
have to be a limit.
|
carson
|
|
response 7 of 239:
|
Jan 7 14:41 UTC 1996 |
Does Grex really offer enough now to keep that many people around?
Are they going to start flocking in simply because Grex gets a little
faster? Grex isn't really that much slower than M-Net over the net;
I tend to find Grex quicker at any given moment.
That's not to say that I don't share Greg's concern over the dangers of
a bloated system. I just wonder if he's exaggerating the impending
danger a tad. For example, nether.net is on a T1 (if not a T3; I forget)
and seems to handle it fairly well, considering Jared doesn't have a
problem with people telnetting in and then right back out. Will putting
Grex on a T1 really max out its staff? I'm not in the best position to
answer.
|
jazz
|
|
response 8 of 239:
|
Jan 7 16:53 UTC 1996 |
Nether's a special case, though. It's directly off of one of the main
local routers for CICNet, has it's own router, and has three x86 boxes
currently running within the domain. It's also a special deal, considering
that Jared is an employee of CICNet.
Even if GREX got a similar connection to ICNet, it wouldn't begin to
compete with the connection situation Nether has.
Then again, Nether has little or no management, no real BBS to speak
of, and seems to be used mostly as an IRC point.
|
srw
|
|
response 9 of 239:
|
Jan 7 18:33 UTC 1996 |
India is a fascinating country. It has 900 million people. A smaller
percentage have access to high tech goodies than in the US, but their
economic potential as a world power has been stifled ever since their
independence by the government's policies. We are finally seeing huge
growth in that country because this has changed, and they have opened up their
economy. Internet activity from India is just a reflection of the
enormous and sudden surge of their economy and access to technology.
I suspect many of them might consider using the conferences if they were not
a little timid about entering such a venue. I don't know what it takes
to convince them to try it, but I hope someone can figure this out.
The irony is that it is probably language, and yet nearly all of them
will have no problems with written English at all.
On friend of mine in Bangalore told me that India's population growth
annually exceeds the entire population of Australia.
|
scg
|
|
response 10 of 239:
|
Jan 7 19:20 UTC 1996 |
(Country 312, if it exists, isn't considered important enough to have its
country code listed in the Ann Arbor phone book)
|
rcurl
|
|
response 11 of 239:
|
Jan 7 20:01 UTC 1996 |
Its not in the 1994 World Almanac either - they have more than the phone book
does.
|
ajax
|
|
response 12 of 239:
|
Jan 7 20:31 UTC 1996 |
Sure it wasn't an area code? That would place it as Chicago.
|
steve
|
|
response 13 of 239:
|
Jan 7 21:59 UTC 1996 |
Re #1: I suppose we could find out where the people were coming
from, if we really looked at the data. But again, it will have gaps
when people supply only the city, etc.
Re: #2: no, this doesn't indicate at all what they do here once
they're on. We can't really, without taking the process accounting
records to some other machine and write come code to analyze whats
going on (Grex doesn't have the spare CPU for this). We can determine
who has conference files, and we can determine who's made responses
in the conferences.
|
popcorn
|
|
response 14 of 239:
|
Jan 7 22:58 UTC 1996 |
You can also look for a .partymsg file to see who has been in party.
And a Mail or mail subdirectory, or an mbox file, indicates mail usage
(though users can use mail without leaving those behind).
|
robh
|
|
response 15 of 239:
|
Jan 7 23:15 UTC 1996 |
I doubt any of our serious mail users would never have had any
of those files in their directories, though a conscientious
ucb-mail user could always delete their mail, or just delete
the mbox file periodically.
I do have a listing of all Grex users with www directories.
The total is 370 users. You'd have to look at our Web stats
page using Lynx (under the "Info About Grex" page) to see
which pages are accessed more often.
Reluctant though I am to say this, if our being on the Web is
really having a big impact on our bandwidth, we should consider
(ugh, do I have to say it? can someone else say it first?
please?) charging a dollar or two per month for having a Web
directory. If we really need to limit the httpd traffic we
get, that's one way of doing it. For the same reason we don't
hand out Internet access to everyone.
A better short-term solution (IMHO) is charging lower rates
for students, since so many of our non-paying users are students.
I know that when I was in high school, $6 a month would be a major
amount of money for me, while (say) $3 a month would be reasonable.
Especially if Mom and Dad were footing the bill.
|
janc
|
|
response 16 of 239:
|
Jan 7 23:58 UTC 1996 |
We could do something like charging $5 a year for non-members to have a web
page. (They'd be includined in the member price). I don't think I'd want to
charge much more than that, and collecting such a low fee monthly would be
silly.
|
robh
|
|
response 17 of 239:
|
Jan 8 00:43 UTC 1996 |
Actually, I like that idea better, janc. It would let our
regular users keep their pages available, and discourage
the non-Grex person from creating an account here just to
have a Web page.
|
steve
|
|
response 18 of 239:
|
Jan 8 00:56 UTC 1996 |
Aside from the probably negative feelings about charging for
www access in the first place, only American users could send us
a check for $5. Anyone from Europe or Asia would have to spend
more than $5 to get $5 to us.
I'm not at all sure that students would 'flock' in any amount
at all to join at $3/month. As my son Damon once pointed out to
me, $6/month was finding 2 pop bottles a day (Michigan having a
$0.10 per bottle/can deposit). Given that inflation has risen
by better than 15% since 1991, I just don't think that the $6/mo
is a barrier to many, at least here in America.
|
robh
|
|
response 19 of 239:
|
Jan 8 01:47 UTC 1996 |
<robh imagines high school children all over Michigan rooting through
public trash cans in the hope of finding two bottles a day to support
their Grex habit>
|
jazz
|
|
response 20 of 239:
|
Jan 8 02:19 UTC 1996 |
It's a standard system administration tool with some Californian public
access systems to put a small script shell around utilities whose use was in
doubt to record the number of times they were used. It was used primarily
to see if a certain group of utilities was worth keeping on a limited amount
of drive-space, but there's no reason it couldn't be turned around and used
to monitor what gets used, when, and how long, and if it outputted a database
format, or was imported into a database, you could even make pretty charts
of peak usage hours, average time of usage, system load at time of usage, et
cetera ... might be an idea when considering what options to spend money on,
if the vox populus is a strong one.
|
steve
|
|
response 21 of 239:
|
Jan 8 02:32 UTC 1996 |
We already have that, courtesy of SunOS. There is a system called
process accounting that records all this info. We wind up with these
multi-meg files every morning, that contain the previous days data.
We certainly can go over stuff, and have in the past.
|
kaplan
|
|
response 22 of 239:
|
Jan 8 02:53 UTC 1996 |
re 5: Yes, according to /u/moo/.plan, moo registered in Dec 93 at speed
14400. But I remember that when I registered at about the same time, grex
had no direct Internet connection. And for a while after the net
connection was established, only specific sites were permitted to telnet
in. So I ask again when anyone was permitted to telnet in. I'm sure it
was later than that.
|
steve
|
|
response 23 of 239:
|
Jan 8 03:46 UTC 1996 |
March 14th 1994 was when the world discovered about Grex, courtesy
of Valerie getting the MSU gopher put us on their list of telnetablle
sites.
In mid February we opened things up, I think. We have all the data
and this is just useless enough a quesstion to warrent going through
it to find out. ;-)
|
mdw
|
|
response 24 of 239:
|
Jan 8 07:33 UTC 1996 |
The hard part about the pacct records is not the data, but the
interpretation. For instance, computer conferencing is a relatively
efficient use of grex, in terms of CPU, disk, & network bandwidth.
Pine, on the other hand, is bad in terms of CPU. Some people have bbs
as their login shell - if they run mail from bbs, they're actually doing
mail, but they also rack up elapsed "bbs" time.
Nevertheless, I think if we investigate the use of the system, the
this is going to show up at the top:
mail - 2 kinds:
mailing lists
interpersonal or local mail
other uses:
ntalk -(the indians like this)
conferencing
lynx & local browing activity
inbound web page service
party
write, tel, & etc.
|
