|
Grex > Coop12 > #14: Internet Connectivity Revisited |  |
|
| Author |
Message |
| 25 new of 176 responses total. |
jep
|
|
response 27 of 176:
|
Apr 20 15:14 UTC 2001 |
M-Net had 4 available inbound lines at first when they set up with
WWNet, but reduced them recently.
re #22: M-Net's needs are different primarily *because* it's financial
situation is different. There was a time when M-Net had a lot of cash,
but that changed through unnecessarily spending some of it, and losing
contributions (partly as a result of bad spending), and for a lot of
other reasons. Fortunately for Grex, none of these things can ever
happen *here* and so Grex doesn't need to look at the possibility of
cutting it's expenses by 50-75% while dramatically improving it's
Internet connectivity.
By the way, log into M-Net and Grex in two separate windows, using their
GUI interfaces, and do a conference listing sometime to compare speeds.
You can do quite a lot of conferencing on M-Net while you're waiting for
Grex to just list the items in a conference. Try entering an item or
response on Grex, and see how many responses or items you can get
entered on M-Net while you're waiting for Grex to save one.
Grex's Backtalk interface is much nicer than WebYAPP. It'd probably be
really nice if it ran as quickly.
|
carson
|
|
response 28 of 176:
|
Apr 20 15:43 UTC 2001 |
(John's technically right. I can whip right through conferences on
M-Net because there's almost nothing worth reading.)
(that's not to say expenses wouldn't be cut by co-location. Grex
currently spends how much on the Pumpkin? $25/month? $35?)
|
scg
|
|
response 29 of 176:
|
Apr 20 15:49 UTC 2001 |
Grex's Pumpkin rent is more than that. I forget how much more. It's also
important to consider power costs, which are sometimes included in colocation
arrangemenets.
|
carson
|
|
response 30 of 176:
|
Apr 20 15:56 UTC 2001 |
(it *would* be a good idea to know what expenses might be saved by a
co-location agreement. phone lines might be the first expense to go.)
|
jp2
|
|
response 31 of 176:
|
Apr 20 16:03 UTC 2001 |
This response has been erased.
|
aruba
|
|
response 32 of 176:
|
Apr 20 16:44 UTC 2001 |
Re #28-29: Grex pays $69.46 per month to rent the Pumpkin, and $54.34 per
month for electricity.
|
gull
|
|
response 33 of 176:
|
Apr 21 03:39 UTC 2001 |
Re #31: Newer doesn't always equal better, though. How does MNet's user
load compare to Grex's? In my admittedly limited experience, newer
Intel PC-class hardware doesn't quite measure up to older Sun
server-class hardware when it comes to handling lots of users. The
Intel boxes just don't seem to handle multiple processes nearly as
well, and the hardware seems to be flimsier and more prone to
deterioration as well. There's more to system speed than how fast the
clock runs.
|
jep
|
|
response 34 of 176:
|
Apr 21 03:44 UTC 2001 |
re #28: I agree to some extent that Grex has better content. It also
has a better WWW interface, more mature and interesting user base,
better management, a better set of rules, and is more stable.
M-Net's Internet connection is far superior to Grex's, and it's hardware
runs faster than Grex's. M-Net's hardware is newer, I think it would be
replaceable off the shelf for less than Grex's would be replaceable for
used, and M-Net has more disk space. (M-Net is unlikely to have any
kind of crisis based on lack of disk space for quite a while, and
wouldn't if it maintained Grex's entire user base -- for that
matter, Grex's entire set of filesystems.)
Carson, you aren't going to insult me by berating M-Net for what it is
and what it isn't. I know that a lot better than you do. I log on to
M-Net every day. (Grex, too.) There *is*, by the way, stuff on M-Net
that's interesting -- if you're not too busy attacking it to look and
see.
|
gull
|
|
response 35 of 176:
|
Apr 21 03:47 UTC 2001 |
I just think it'd be a major expense for Grex to replace the current
hardware with something that'd be both faster and more reliable. A lot
of PC hardware (especially hard disk drives) is cheaply made for the
desktop market and would fail quickly under the kind of load Grex would
put on it. You're talking server-class multiprocessor stuff,
realistically, I would think, and that doesn't come cheap.
|
jep
|
|
response 36 of 176:
|
Apr 21 04:13 UTC 2001 |
I've heard the arguments like that ("server class" and all that), but
outages such as last summer's several weeks of downtime aren't because
of hardware problems; that was because of a hacker and a system upgrade.
Other than that, M-Net's been just as reliable as Grex over the last
decade, during which time M-Net has been running on PC hardware.
Specifically, M-Net hasn't had any disk failures that I've been aware
of, not in 10 years, but I believe Grex has had them.
|
carson
|
|
response 37 of 176:
|
Apr 21 08:36 UTC 2001 |
resp:34
(I wouldn't go so far as to say that you're that much aware of what's
going on re: M-Net than I am. having kept myself abreast of the system
ever since the Salcedo incident, I'm one of the few people you *can't*
accuse of not being familiar with both M-Net and Grex. plus, I've held
a "position of responsibility" on M-Net more recently than you have.)
(if you think my comment about M-Net not having much worth reading was
meant to offend you, so be it. the fact is that M-Net has newer
hardware, a faster net connection, and more disk space, all for naught.
it's ready for a big boom that isn't likely to happen.)
(seriously, why bother pointing out M-Net? *Chinet* [which has been
around even longer than any permutation of M-Net has] has newer hardware
than M-Net does, will have its own T-1 in a matter of days, and has
unreasonable amounts of disk space [mostly used for games]. it also has
maybe 12 users. I can post an item or enter a response there very
quickly; the wait is in a response from someone else.)
(here's my theory: newer hardware, faster net connections, and oodles of
disk space have *zilch* to do with building a community. all that speed
is only good for allowing inane users to post inane responses insanely
fast. seriously, what's the speed supposed to be used for? people can
only read and type so fast. whether you're connected to Grex by phone or
by port, it's usually at the speed of type or faster. the only difference
is when using Backtalk, and that's mostly in retrieval, and *still* as
fast as any database I use on the internet. ever tried to do a search on
cnnsi.com?)
(I'm reminded of janc's comments on why he doesn't speed when he travels:
at some point, the increase in speed isn't worth the time gained. think
of it as diminishing returns.)
(you're patently wrong about the ease of replacing hardware. last I
checked, Grex had plenty of spare CPUs laying around the Pumpkin, ready
to go. M-Net, as last summer showed, obviously doesn't. which would
be cheaper to replace, again?)
(I'd also be curious to see how well M-Net's current set-up might handle
the sort of traffic Grex sees on a regular basis. I'm unlikely to see
that experiment anytime soon, as M-Net has pissed away its userbase and
seems xenophobic about expanding it. [*cough*allocatedtelnetports*cough]
that's fine by me. while M-Net continues to attract users that detract
from the system, I'm happy to suffer through a few thousand script-kiddies
if it means I'll see a few more users on a regular basis. it weren't for
Chinet, M-Net could be the fastest system no one uses.)
(I'm all for a discussion of expenses and cost-saving measures, but I'd
hate to see it focus on "improving" a resource that won't necessarily
produce a benefit.)
|
keesan
|
|
response 38 of 176:
|
Apr 21 15:07 UTC 2001 |
Lynx runs about twice as fast on mnet as it does on grex, and the modems are
also faster. I don't see how this affects the conferences, but it is nice
when you are browsing or downloading files. The conferences are used by
different people for different purposes (more for socializing than for
exchanging information). Obviously the M-Net users think that the responses
on M-Net are worth reading and many of them don't think the same for grex.
I don't see the point of name-calling. I hope both systems continue to exist.
|
jp2
|
|
response 39 of 176:
|
Apr 21 18:13 UTC 2001 |
This response has been erased.
|
jp2
|
|
response 40 of 176:
|
Apr 21 18:22 UTC 2001 |
This response has been erased.
|
scg
|
|
response 41 of 176:
|
Apr 21 23:53 UTC 2001 |
Grex's network connectivity shouldn't be measurably worse than M-Net's. DSL
connections are pretty low latency, and WWNet and Voyager both have
connectivity to the rest of the Net that's decent but could use some
improvement. I think Voyager's is somewhat better than WWNet's at the moment.
Traceroutes from M-Net to Grex currently go through New York, so neither of
them is buying connectivity from providers that have very good peering with
eachother. If M-Net's connectivity is considerably better than Grex, it's
more likely because Grex's DSL line is filling up than anything else.
|
jp2
|
|
response 42 of 176:
|
Apr 22 01:00 UTC 2001 |
This response has been erased.
|
jep
|
|
response 43 of 176:
|
Apr 26 14:02 UTC 2001 |
Using Backtalk, it does take a very long time (30-60 seconds) to go from
one item to the next, or to post a response or a new item. Resp #37
refers to diminishing returns for speed. I start doing something on
Grex every day which takes too long, and so I go to do something else,
and then the thing I was doing doesn't get done. This isn't a trivial
problem, and I don't think it's an unsolvable one.
|
jp2
|
|
response 44 of 176:
|
Apr 26 17:45 UTC 2001 |
This response has been erased.
|
carson
|
|
response 45 of 176:
|
Apr 27 00:32 UTC 2001 |
resp:44
(you're joking, right?)
|
jp2
|
|
response 46 of 176:
|
Apr 27 02:10 UTC 2001 |
This response has been erased.
|
gull
|
|
response 47 of 176:
|
Apr 27 02:38 UTC 2001 |
It doesn't normally take 30 seconds for me to move from one item to
another in Backtalk. The speed of web browsing on Grex is about what
I've come to expect from other sites. Maybe I'm just too used to my
28.8 modem connection, and when I've had more bandwidth for a while I'll
start to get annoyed with sites that don't load instantaneously.
|
scott
|
|
response 48 of 176:
|
Apr 27 11:10 UTC 2001 |
With Backtalk you're probably also seeing processing delays.
|
carson
|
|
response 49 of 176:
|
Apr 27 12:55 UTC 2001 |
resp:46
(from Arbornet's "support" command):
Patron Member Guest
Inbound telnet ports 64 56 48 from Michigan,
28 otherwise
(again, you're joking, right?)
|
jp2
|
|
response 50 of 176:
|
Apr 27 13:58 UTC 2001 |
This response has been erased.
|
carson
|
|
response 51 of 176:
|
Apr 27 15:06 UTC 2001 |
resp:50
("As a current Board member and former staff member, I am pretty
convinced I know more about what is going on that you do?" that's
just *too* easy.) :P
(again, you're joking, right? I haven't even addressed site-bans
yet.) ;)
|