Same problem for over 6 months.

@k2gremlin

 Your picture didn't display but I did look at your connection from this end and all seems to be as it should.  Could you try signing into your http://www.cox.com account and do some speed tests with and without the router?

Chris, for the last 6 months or so I have had technicians out many times. I have hooked directly to the modem and have had the same results. I am running a speed test right now from my router. The fact that I get great speeds throughout the day from the device connected to the modem and then crappy speeds during peak hours has only proven one thing to both me and the technicians in my area. In fact, one of the technicians that regularly comes to my house lives a few blocks away and says he has the same problems.  I ran this latest test from a CentOS machine that is directly connected to the modem. Here are the results. 

Retrieving speedtest.net configuration...
Retrieving speedtest.net server list...
Testing from Cox Communications (70.179.146.20)...
Selecting best server based on latency...
Hosted by Wamego Telephone Company (Wamego, KS) [25.47 km]: 34.917 ms
Testing download speed........................................
Download: 8.09 Mbit/s
Testing upload speed..................................................
Upload: 11.36 Mbit/s

8 Megabits.. that is 8% of what I pay for. Garbage. This CentOS is connected via 1gig ethernet and works fantastic during the day.

With all of that being said, one of the regional managers for my area has specifically told me that a "node split" or some other term was in the process of being done. All I am looking for is some concrete dates on when this will be accomplished. The reson for the split is that apparently the node referenced in my image of 10.37.36.1 is over-loaded. To paraphrase him, "The area has overgrown the node" meaning that there are too many people connected to this specific node. Peek hours they have too much traffic. 

However, as a customer, this should not matter to me but it does. What he is basically telling me is that they advertised X speeds and cannot support even 10% of that. If you sell 100Mbit connection to say 10 people, the node should be able to support 1000Mbits right? You sell 100Mbits to 100 customers, then the node should be able to handle 10000Mbits. Cox has oversold the node and I want the issue resolved as fast as possible. Truth be told, my home uses the internet a lot. My wife watches a lot of her shows or whatever she watches via Hulu/Netflix. I on the other hand play some online games. However, during peek hours, if she decides to watch a show I am screwed. With a max of around 8-15Mbits with NO load on our network, a single show streamed on Hulu will drop the average speeds to below 5Mbits and we start seeing periods of buffering. Could she reduce the quality setting of the show she is watching? Sure, she could... but we pay for a connection so that we shouldn't need to do this. On the gaming side, when this node gets overloaded, latency goes up. As anyone on here can agree with me, in an online game, latency is critical. During the day and even after 11PM or so when this node is working great, I average around 30ms latency in most games I play. During peek hours, I average over 200ms and consistently hit and hover around 400ms. Unacceptable in my opinion. 

Sorry if I come off a bit technical but networking is what I do for a living. I do not claim to know it all and have learned many things from the frequent Cox visitors that have come to my house to tell me that the equipment in my house is good and that the problem is further down the line. I really enjoy when they come to visit because they see exactly what  I see.

Chris, I know you do some of your own tests. Could you do me a favor and do them again between the hours of 7PM and 10PM CST. You running a test at 5AM when all of the subdivisions in this area are asleep will only show a perfectly clean connection. I agree with you that at 5AM everything seems to be as it should. Just test during peek hours and you will see the difference. 

Thank you,

Matt

Here is a test done from mid morning with 2 kids doing whatever 2 kids do on the internet on Sunday mornings..

Retrieving speedtest.net configuration...
Retrieving speedtest.net server list...
Testing from Cox Communications (70.179.146.20)...
Selecting best server based on latency...
Hosted by Wamego Telephone Company (Wamego, KS) [25.47 km]: 32.09 ms
Testing download speed........................................
Download: 96.01 Mbit/s
Testing upload speed..................................................
Upload: 11.45 Mbit/s

Chris, the problem is that the saturation of the node happens in the evening hours every single day without fail. The connection is not intermittent, it's just slow during peek hours. I don't know if it is just a limitation on the physical cable, processors, or memory on this node.. but whatever it is, I sure hope it is resolved soon.

Matt 

In all the info in the post, none of the important stuff is mentioned.

1. What modem/router/gateway?

2. What are your signal levels? (When the problem is and is NOT occurring)

3. Can you connect direct to the modem and try the trace again when the problem is occurring??

4. Last, try this speed test.

Modem:DPQ3212

Switch my PC is connected to is a 3560G

Router is a CentOS VM running on an R710. 

As you can see from my history of using the cox speed test on their website, peek hours is horrible.  These results are both with and without a direct connection to the modem. 

The point that I am trying to make is that the problem has been FULLY recognized by the technicians that come to my house, local field managers, as well as regional managers. The node is over populated and will be split "soon" according to them. 

Let me put this is caps to be more clear. ALL I AM LOOKING FOR IS A CONCRETE DATE FOR THE INSTALLATION OF THE NEW NODE  TO REDUCE THE STRAIN ON THE CURRENT NODE. The problem has already been identified and confirmed on many levels. 

You do networking for a living and that's the best network diagram you can do? DQP3212 > Switch > Router? If so, why? More then 1 public IP? Is this a residential account? 

Also, I don't "think" I saw you mention bypassing the router and switch and testing that way. Your first tracert show 300ms+ from CPE to CMTS, yet your speed tests show latency as low as 12ms. Is the latency location specific? Time specific? Need more data. What are you signal levels? Etc

As for getting a concrete date for a new node, I think you are barking up the wrong tree. This is mostly a user forum, with moderators that work for Cox's social media team. There are no engineers here. Even if there were, THEY wouldn't be able to give you a concrete data. I have personally seen tickets for node reworks, and usually the time spread is 3-4 weeks, between prep work and construction. You might try DSLReports and see if you can PM one of the engineers there. Or buy a beer  for one of the COX maintenance guy if you see him on the street. THEY would be the best ones to know. But don't expect a exact data. That is unrealistic IMO.

Sorry if I came off as harsh. But as someone who is also a network engineer, I expect more from my fellow nerd. 

Here is a network diagram that may suit your visualization needs better.

I listed the equipment earlier in the wrong order and I am sorry for any confusion that may have caused. Yes this is a residential account. As it is right now, I can run pings, trace routes and the like directly from the CentOS VM. When the problem hits again tonight, I will pull the plug on that guy and connect my desktop directly to the modem and run the same tests I have ran for the last 6 months. I will log and screenshot everything I can.

Like you said, a typical node rework should only take 3-4 weeks. When this problem first started, it took almost 2 months to get Cox to isolate the issue to the node. Since then, I have been told many stories on when this re-work will happen. First it was sometime in the 3rd quarter of last year. When that time frame passed, it was sometime in the 4th quarter. Here we are, almost through the 1st quarter of this year and still no fix has been implemented.

I guess in a way I am using this post as a way to start logging everything that happens in regards to the current problem.

Here is the list of ticket numbers I have accumulated over the last 6 months. Maybe someone here has access to these tickets and can provide more information.

71827223 - Tier 1 or Tier 2 support ticket.

1369076 - Escalation ticket.

2783451 - Tier 2 ticket.

9030163 - Tier 2 ticket.

I will make my usual calls around tomorrow in hopes of getting more information. Otherwise, I will just continue to post my frustrations here in the hopes that someone in customer support or upper management sees this and decides to take action.

k2gremlin said:
Here is a network diagram that may suit your visualization needs better

Thank you, I actually am very visually focused, so it helps. Looks like a nice setup. Kind of curious what gear your running your CentOS but thats a question for another thread.

If you connect direct to your modem and see anything higher then 10-30ms to the first hop (depending on your area) that is strong evidence of a issue with signal levels like yours.I just don't see why congestion would cause latency spikes that bad unless it was REALLLLLLLY congested. Like 1000+ on one node kind of congested. Typical is 300-400. I think something else is going on too, but I don't know what. 

With your background and the technical complication of this issue might sit better over on DSLR's Cox Forum. The moderators here can contact some of the engineers found on that site, but sometimes it's easier to cut out the middle man/women. See here for a list of official employees. Start a thread and PM one of them (in your market) the personal details like the ticket number and MAC of your modem. You can even keep this thread open and cross reference and fact check.  

FYI, I closed my DSLR account due to a conflict with one of the mods, so I won't be able to see your thread there.

Teck,

Thank you for the heads on on the DSLR forums. I will be posting there shortly. Here is the connected statistics. Having a hard time getting images to post correctly. If you right click and open in new tab you can see the entire picture. 

Was the tracert, ping, and speed test all done at around the same time?

What makes me curious is why I see huge 500+ms spikes there, but they don't reflect in the tracert or latency read out on the speed test. Just bad timing? 

Yea, this is a tough one. I fear it's going to get messy between you and Cox. If what you suspect is true, and they let a node get THAT congested, there is probably some town/local bureaucracy that is holding up the process. Did they at least say when the new node will be built? Like have you seen any Cox or contractors installing new stuff anywhere? Once built its usually only a couple weeks before the node split, so at-least that would give you some hope. If not, I would first try DSLR and if that doesn't work, ask for either free service or no service and find another provider. 

Also try to work with Tier 2 or Tier 2.5 (if it still exists) and the Customer Advocate Department. Get ticket numbers, and as much data as possible. Even if it doesn't fix the problem, it would be good ammo for FCC/BBB/ETF fight. I would offer to ask some of my friends but I burned up all my favors trying to get my own node problem fixed. :-( Good luck though.