dslreports logo
site
 
    All Forums Hot Topics Gallery
spc

spacer




how-to block ads


Search Topic:
uniqs
1453
share rss forum feed

lancorp

join:2009-11-15

[CA] Service quality issues that COX San Diego can't fix

After having dealt with COX Business San Diego for over a month on this issue without resolve, I thought I would get some opinions here (possibly to help COX figure it out).

It started a couple of months ago. As an IT person, I keep many gadgets on my desktop. One is a ping monitor that monitors ping times to various hosts (I use 3 or 4 different DNS servers). All of the sudden, instead of 10-40ms ping times, I was getting large sweeps of up to 1000ms and lost pings. I reset my cable modem and the problem went away. FOR A WHILE...

More and more frequently, I started having the same problem. All DNS server ping times would be extremely long and packet loss would be 10% or more. Power cycling the cable modem would always fix it. These last few weeks, it happens almost every day. I've called CBS (I have residential business SOHO service), and they've checked this and that remotely. Once, they confirmed others on my node appeared to have high latency also. A couple of times, they have sent out a tech to troubleshoot, but of course, when the tech arrives, the internet is working normally. The original Cisco modem was replaced with a Moto 6180, but same symptoms, so I doubt it's the modem.

A symptom of the problem is a bit weird...when the latency is high and there is packet loss, using more download bandwidth causes the latency to improve greatly. Stop the download (stream video, VOIP call, etc.) and the high latency immediately returns. On a VOIP call once to COX during the problem, I could hear him perfectly (downstream) but he said I was very choppy (upstream).

Last service tech call, the tech was a higher level tech. He was in charge of a recent recable in our neighborhood, and knew everything about our area. It hooked up his laptop and checked everything. Signal levels are perfect. No sign of problems.

Last couple of days, even during "normal" periods when ping time are good, we've been getting messages from various devices that our connection was too slow (amazon instant video thru TV and downloading Office 365 via PC early in the morning). After doing about a dozen tests at speedtest.net, it was showing the speedtest results were very slow down, and superfast up. Down would be 2-5 and up would be 15-25. We have 25/5. The graph that speedtest.net shows during the test would be peaked for the first half the graph, then, about in the middle, would drop to the bottom for the rest of the test. Every once in a while, the speedtest would show normal, but mostly would be superfast for half then drop to practically nothing. I tried these tests on 4 different PC's, wired and wireless in my house. I also connected my laptop directly to my cable modem, reset everything, and ran the speedtest with similar results.

I'll probably try calling them again this morning, but am looking for tips, clues and help on what the problem could be. This has been a long, on-going issue and I hope I haven't left anything out, but if something above sounds incomplete, I'd be happy to fill in any blanks!

Thanks.

intel1212

join:2008-01-17
55134
Reviews:
·VOIPO
·Cox HSI
Not sure what kind of network you have, if you have validated you don't have a local type of problem, but how do you know that you don't have cabling issues within your local area network? I use a tool from time to time to get a sense of "quality" of my home network, which consists of both WiFi and Cat 5e. Tool is Ping Tester. I like it because I can set time value between packets (very short) and a fixed amount of packets (like 2K). It acts like a load testing tool.

If you haven't validated your local network, this info would help isolate what the problem is or isn't.

Hope this helps.

Smith
Premium
join:2000-07-13
reply to lancorp
What kind of ping/speed test results do you get using Cox's 'Speed Check' tool?

lancorp

join:2009-11-15
reply to intel1212
I have an extensive gigabit network at home. I have done what I could to rule out out LAN issues.

With the issues of high latency and dropped pings, simply resetting the cable modem fixes it. That would not (to me) point to a LAN issue. Early this morning, I connected my laptop directly to the cable modem to witness the slow download speeds, so that would rule out a LAN issue, also? Unfortunately, in order to disconnect my LAN from the modem and connect a laptop directly seems to require a modem reset in order for the latop to DHCP an IP address from COX. In doing so, if I am having latency issues, the modem reset itself clears the problem and I am not able to test.

lancorp

join:2009-11-15
reply to Smith
said by Smith:

What kind of ping/speed test results do you get using Cox's 'Speed Check' tool?

Where do I find a COX speed check tool for San Diego?

Smith
Premium
join:2000-07-13
On Cox's website, go to: My Connection > Internet Tools > Speed Check

intel1212

join:2008-01-17
55134
Good idea. I too find speedtest.net vs. Cox Speed Test show differing results, mostly on the download.

lancorp

join:2009-11-15
reply to lancorp
Just to follow up on this...COX found that a neighbor was injecting noise into the system and COX says that was causing the problem.

Supposedly, they've taken care of the noise issue, and the past two days have been good.

Time will tell.


avmar

@fincomfg.com
I have to chime in here for a recent CBS experience similar to what's being discussed. Wichita, KS. Three different CBS businesses each a few blocks apart from each other. Cox data plans 10/2, 10/2 and a 5/1. Small 6 computer network, Cox replaced all coax to the pole, new Docis3 cablemodem, I installed new Cisco 5505 router. Repeatable behavior is: from an workstation start a ping to 8.8.8.8 and from an outside source start a ping to the public IP of the router. Observe normal 20ms-60ms ping latency. From any workstation start an "upload" to the internet of any kind..ie.. the upload portion of a "speedtest", or an ftp upload or anything that sends a substantial file upstream. Now observer the ping latency jump from 20ms-60ms to 150ms-250ms. That's just one workstation sending one file upstream. Kill the upload and the latency immediately returns to 20ms-60ms. Just think if there were several uploads to dropbox or something similar. I do not think it was like this a few weeks ago but cannot prove it. Cox claims that this is normal behavior. Has anybody else seen this? Just curious. For testing purposes Cox has increased the datan plan to 24/4. New behavior is that one upload connection does not affect the ping latency, but a second upload connection does just as before. As one can imagine, vpn tunnels and rdp connections do not work well with connection latencies in the hundreds of ms.


Optimus2357
Premium
join:2010-11-21
West Warwick, RI
kudos:3
Your problem sounds different then his/hers so you may want to make a new post so it gets the attention it deserves, but off the top of my head I would look into QoS or some kind of traffic throttling you have on the Upload of the WAN. Even if there is no other upload bandwidth except yours the router itself may be creating the latency. See »www.cisco.com/en/US/docs/securit···p1083751

Or you might be saturating your upload and causing latency at the modem. I say this because the problem decrease after increasing your upload. I would say bypass router to isolate first and then open some heavy traffic and see if the problem occurs. If it doesn't I would look again at your WAN settings and QoS and disable it all together if possible. If you do see the problem after bypassing I would look to something occurring with any modem in that spot(EMI, over heating, less then full wattage going to the power brick, etc). Maybe try a different Coaxial outlet? JMHO


avmar

@yodersmokers.com
reply to lancorp
Thanks for the reply. I will start a new thread. Also thanks for the link, that's certainly something to look in to. I've just never had to explicitly put Qos in place on small accounts like this, their upload traffic seemed to managed itself and not hog the outbound connection.