As a subscriber to Comcast's cable internet service in the Magnoila
neighborhood of Seattle, I have been suffering from poor connectivity for
about 4 months. I theorize that I am losing connectivity for brief periods
(a second or two) every 3-5 minutes. This is a mere annoyance in
"dropout-tolerant" applications like web browsing or streaming media feeds,
but it is unacceptable for online gaming, VPN client, and similar
"dropout-intolerant" applications.

I am having problems diagnosing and proving the problem. Pinging Google and
Yahoo websites seems to be AT&T's standard for measuring the quality of my
connectivity - but pinging Google and Yahoo are rock-solid for me all of the
time. I leave ping on continuously sometimes, but even when I experience
dropout, ping keeps faithfully returning packets with no appreciable change
in latency. For example, I connect to a streaming media feed from
http://www.kuow.com/, and ping it's IP continuously. My media feed will
falter regularly, but pinging that IP never falters. Even Internet Explorer
pauses regularly to "reaquire" a TCP connection. As you would suspect from
the above, my modem never "loses sync" with network.

I feel that I am using the wrong means to monitor my network conditions.
Can anyone here give me any insights into how I can observe my connectivity
in a meanginful way, given the intermittent nature of it?

DT
WinXP SP1

"Dirty Tleilaxu" <dt@tleilax.net> wrote in message
news:6nTDb.375428$Dw6.1222791@attbi_s02...


www.pingplotter.com

"Dirty Tleilaxu" <dt@tleilax.net> wrote in
news:6nTDb.375428$Dw6.1222791@attbi_s02:

I feel that I am using the wrong means to monitor my network
network monitor software. Way back when I was a customer, I used to print
my lack of service out and hand it to them (see below) at the office when
they tried to tell me there's nothing wrong with any Comcrap service.

I saved a few to disk, just for old time sake. Here's what mine used to
look like before I came to my senses and went with Knology, the new boys in
town with a NEW, not old cluged up one bought from someone else, network.

2000-08-23 15:01:41 Cable6/0/U3 10.4.67.238 3 online 27.56 39522121
3415190958 72198807 1258139540 0 18795 9927 5 98
2000-08-23 15:17:42 Cable6/0/U3 10.4.67.238 3 online 27.14 39522124
3415191150 72198809 1258139668 0 18834 9927 5 98
2000-08-23 15:33:42 Cable6/0/U3 10.4.67.238 3 online 27.11 39522124
3415191150 72198809 1258139668 0 18873 9927 5 98
2000-08-23 15:49:42 Cable6/0/U3 10.4.67.238 3 online 26.95 39522127
3415191342 72198811 1258139796 0 18911 9927 5 98
2000-08-23 16:05:43 Cable6/0/U3 10.4.67.238 3 online 23.90 39522130
3415191534 72198813 1258139924 0 18950 9928 5 98
2000-08-23 16:21:44 Cable6/0/U3 10.4.67.238 3 online 24.01 39522130
3415191534 72198813 1258139924 0 18989 9929 5 98
2000-08-23 16:37:45 Cable6/0/U3 10.4.67.238 3 online 18.43 39522133
3415191726 72198815 1258140052 0 19027 9930 5 99
2000-08-23 16:54:14 Cable6/0/U3 10.4.67.238 3 online 14.52 39522133
3415191726 72198816 1258140665 0 19067 9962 5 99
2000-08-23 17:10:15 Cable6/0/U3 10.4.67.238 3 online 19.61 39522134
3415191790 72198817 1258140729 0 19106 9991 5 100
2000-08-23 17:26:16 Cable6/0/U3 10.4.67.238 3 online 15.64 39522134
3415191790 72198817 1258140729 0 19144 9995 5 100
2000-08-23 17:42:17 Cable6/0/U3 10.4.67.238 3 online 17.61 39522137
3415191982 72198819 1258140857 0 19182 9997 5 100
2000-08-23 17:58:18 Cable6/0/U3 10.4.67.238 3 online 19.56 39522137
3415191982 72198819 1258140857 0 19221 9999 5 100
2000-08-23 18:14:18 Cable6/0/U3 10.4.67.238 3 online 20.60 39522140
3415192174 72198821 1258140985 0 19260 9999 5 100
2000-08-23 18:30:19 Cable6/0/U3 10.4.67.238 3 online 19.80 39522140
3415192174 72198821 1258140985 0 19299 9999 5 100
2000-08-23 18:46:50 Cable6/0/U3 10.4.67.238 3 online 21.94 39522143
3415192366 72198823 1258141113 0 19338 9999 5 100
2000-08-23 19:02:50 Cable6/0/U3 10.4.67.238 3 online 25.64 39522143
3415192366 72198823 1258141113 0 19377 9999 5 100
2000-08-23 19:18:51 Cable6/0/U3 10.4.67.238 3 online 27.20 39522146

See the number XX.XX right next to "online"? That was my signal to noise
ratio. Anywhere below 20 db used to produce ping times that looked like
this to the LOCAL NODE.

09/15/00 22:54:27 ping 24.1.32.1
Ping 24.1.32.1 ...
1 Addr:24.1.32.1, RTT: 14ms, TTL: 254
2 Addr:24.1.32.1, RTT: 86ms, TTL: 254
3 failed
4 Addr:24.1.32.1, RTT: 34ms, TTL: 254
5 Addr:24.1.32.1, RTT: 12ms, TTL: 254
6 failed
7 failed
8 failed
9 failed
10 failed
11 failed
12 failed
13 failed
14 Addr:24.1.32.1, RTT: 13ms, TTL: 254
15 Addr:24.1.32.1, RTT: 14ms, TTL: 254
16 failed
17 Addr:24.1.32.1, RTT: 29ms, TTL: 254
18 failed
19 failed
20 Addr:24.1.32.1, RTT: 34ms, TTL: 254
21 failed
22 Addr:24.1.32.1, RTT: 101ms, TTL: 254
23 failed
24 Addr:24.1.32.1, RTT: 34ms, TTL: 254
25 Addr:24.1.32.1, RTT: 35ms, TTL: 254
26 Addr:24.1.32.1, RTT: 81ms, TTL: 254
27 Addr:24.1.32.1, RTT: 14ms, TTL: 254
28 failed
29 Addr:24.1.32.1, RTT: 12ms, TTL: 254
30 failed
31 Addr:24.1.32.1, RTT: 11ms, TTL: 254
32 failed
33 Addr:24.1.32.1, RTT: 11ms, TTL: 254
34 failed
35 Addr:24.1.32.1, RTT: 33ms, TTL: 254
36 failed
37 failed
38 Addr:24.1.32.1, RTT: 13ms, TTL: 254
39 Addr:24.1.32.1, RTT: 13ms, TTL: 254
40 Addr:24.1.32.1, RTT: 12ms, TTL: 254
41 Addr:24.1.32.1, RTT: 12ms, TTL: 254
42 Addr:24.1.32.1, RTT: 14ms, TTL: 254
43 Addr:24.1.32.1, RTT: 13ms, TTL: 254
44 Addr:24.1.32.1, RTT: 14ms, TTL: 254
45 Addr:24.1.32.1, RTT: 12ms, TTL: 254
46 Addr:24.1.32.1, RTT: 13ms, TTL: 254
47 Addr:24.1.32.1, RTT: 119ms, TTL: 254
48 failed
49 failed
50 Addr:24.1.32.1, RTT: 12ms, TTL: 254

Now, they'll tell you that was back in the @Home (we called it @Hosed)
days, but that's not true. This was in the LOCAL system, not the ISP that
went under. I still know people who have times like this.

Every cable operator runs software like this and knows EXACTLY what's going
on inside his plant to EVERY user EVERY few minutes because the software
polls the modem on your desk. See that cable light blinking for no
apparent reason even though the computer is off? Every wonder why?...(c;

Of course, we wouldn't want to ADMIT something is still @Hosed, now, would
we?

Larry

Larry W4CSC <noone@home.com> wrote in
news:Xns94DED48116A04w4csc@216.168.3.44:

@Home dictated what happened at the Local ISPs too. In the case of my
cable ISP, @Home decided the firmware versions to be used for the cable
modem routers. They wouldn't support the ISP unless the ISP followed @Home
"standards" to the letter. The particular firmware in question immediately
caused problems with all UDP type traffic. Packet loss if UDP / ICMP
traffic went through the roof, up to 99% for extended periods of time.

At first, the ISP denied any problem. I had to go to another ISP in
another city, and capture network logs that showed a user in another city
had a better connection to the ISP than their own users in their own city
did. At first the ISP claimed I'd fabricated them, then finally admitted
there was a problem that could be demonstrated easily. But then they said
they could not change the firmware until @Home said they could, which was a
year at earliest.

Eventually some Motorola techs who happened to be avid online gamers as
well got wind of the problem with the firmware. The problem had never
shown up at Motorola because they never test the firmware beyond basic
functionality. They had no test to simulate what happens with 200+ users
on a router. Anyways, they finally reproduced the problem and created a a
firmware fix. But @Home still insisted in not changing the firmware. At
this point anyone still interested in online gaming was switching to the
DSL ISPs in town.

@Home was doing a lot of other things that finally broke the deal between
my cable ISP and themselves. Since we were one of the largest cable modem
plants in north america, the traffic was quite high. @Home would routinely
disconnect our ISP because their system could not handle the traffic. They
would do this without any notification to the ISP. There were supposed to
be redundant paths but @Home never got them working. During this time, my
ISP kept their old connections to the net operational. Eventually they
started using them as backups when @Home would disconnect them, and later
on they just started running on the "backup" connections full time.
Eventually @Home and the ISP went separate ways and at that time the
firmware on the cable routers was upgraded. About 1 year after the problem
was created, it was fixed.

I don't doubt there are still things left behind from the @Home days. For
starters, the people who first ran the operation at my ISP were really
dedicated to the customers, and very technically skilled as well. You were
able to get real answers and support when you needed it. Now support is
mostly people answering the helpdesk phones who don't have a lot of tech
experience. The really skilled senior techs were treated with little
respect or consideration during the @Home phase. It was the @Home way or
the highway. This was hard to take for these people, who had been testing
and developing the cable modem infrastructure since 1996. Their knowledge
and expertise was not appreciated by @Home and so they eventually left the
ISP, rather than watch @Home destroy what they had built.

"Mr. Grinch" <grinch@hatespam.yucky> wrote in
news:Xns94DF36DACC67Bgrinchhatespamyucksh@24.71.22 3.159:

But, @Hosed had nothing to do with the signal to noise ratio, an RF and RFI
issue in the plant at Comcrap. The TV signal had about the same noise
levels in it. But, TV is more tolerant. The plant is OLD, never designed
for upchannel, low frequency signals. Knology, on the other hand, with its
copious UPS/natural gas gensets spread across the city, was designed with
TV, broadband and their telephone-over-cable-plant system in the first
place. So, signals look like:

Downstream Channel
This information display the current signal status of downstream channel.
Lock Status Locked Modulation qam64
Channel ID 1 Symbol rate 5056941 sym/sec
Downstream Frequency 585000000 Hz Max Bit Rate 2048000 bps
Downstream Power 20.63 dBmV SNR 37 dB


Upstream Channel
The information display the current signal status of upstream channel.
Lock Status Locked Modulation QPSK
Channel ID 10 Symbol rate 2560 ksym/sec
Upstream Frequency 24000000 Hz Max Bit Rate 256000 bps
Upstream Power 25.75 dBmV

(our modems aren't locked up to hide what the signals look like.)

37 dB signal to noise ratio makes broadband at 2Mbps impeccable unless
someone wraps a redneck pickup around a phone pole and tears it down. My
modem log was old until the Comcrap installer chopped my underground drop
in TWO places installing TV to a neighbor.

Larry on Knology....2Mbps and no byte limit on 2 ports at 512Kbps/port
Supernews with full binaries downloading 24/7/365.

Larry W4CSC <noone@home.com> wrote in
news:Xns94DF66A8ADDC1w4csc@216.168.3.44:

I don't remember the optimum levels for the Motorola CyberSURFRs, but I know
they are a bit different. Regardless, you are right, levels are up to the
ISP, and many if not all of them are lax in adjusting those levels for the
optimum. If the signal is at the outer edges of the operating range, they
will just leave it rather than send someone out to adjust it. A really hot
day is sometimes all it takes to change a bad cable run enough to push the
signal level outside of the range and cause problems. End result, more
support calls that could have been avoided by just sending someone out and
adjusting the levels in the first place.

In defense of the techs who originally ran the show at my ISP, they always
made an effort to adjust signal levels for optimum, re-running cables
sometimes if necessary. I don't see anyone doing that today.

I'm sorry, I just had to laugh out-loud when I read about the guy cutting
your drop. But there is no monopoly on knuckle draggers, as hard as some
companies seem to try.

I get 10mps downstream and 768kbps up, in theory. In practice, I can do a
peak of 7.5mbs down and maybe 300kbps up. I can get sustained 5mbps down if
I hit multiple news servers. But my max is 50GB a month. I can get things
in a hurry but I also get into trouble with the ISP in a hurry.