DPC3939B slow response time

Hello guentha and welcome,

The next opportunity you have, I recommend you contact 8003913000, option=technical support, and have the technical agent make absolutely sure of the following settings on all your DPC3939B are as follows:

1. WAN DNSs are programmed to Primary = 75.75.75.75 and Secondary = 75.75.76.76

2. Your LAN DHCP Service DNSs are programmed to Primary = 75.75.75.75 and Secondary = 75.75.76.76

3. Your LAN Leasetime=forever

Also, if you are using static IP addresses on any firewall or controlling router, make sure your WAN interfaces have DNSs are programmed to Primary = 75.75.75.75 and Secondary = 75.75.76.76. 

If you are getting 32 or 64 byte pings in 1ms or less, you have a phenonimal network inter-structure. Actually between 10ms to 100 ms is an average range that I have seen with most business class customers. If you are ever measuring packet loss by pinging any DNS servers this may give you negative results due to most DNS servers,like Comcast's for sure, have built-in accessive protection facilities and this will always give you up to 10% packet loss.

Hope this helps you out.

I should preface this post by letting you know I and all my employees are Network Engineers.  

Our internal network is indeed phenomenal. I am blessed with the budget to run cat6a and fiber to higher end Cisco switches but sub 1ms ping times inside a network are not all that hard to achieve with normal switching equipment. We do run layer 3 switches with quite a few rules which does complicate things but we invest in the hardware to make sure it doesn’t cause a problem. Nothing on our network takes even a full millisecond for response.

Your suggestions of DNS and DHCP settings have no bearing on my situation as it’s actually the gateway causing the lag and I am pinging with IP address. I am not even going over Comcast’s WAN (coax network) I am just pinging the gateway IP which should come back in ~1ms. 

When I run this test at several other sites I find it is a crapshoot even with different gateways. Some do respond to pings in 0-1ms some exhibit the behavior I have mentioned with pings sometimes taking up to 200ms to return. This leaves only 2 options. A software problem or a hardware problem as the rest of the network has been ruled out by connecting directly to the gateway and testing with multiple known good cables. 

I have a hunch it’s with the way statics are programed. There seems to be some variance in how it’s done. Having lots of sites to check I have noticed some differences. Like sometimes with a /30 the gateway has the highest IP in the range while at other sites it has the lowest IP in the range. It looks like all my sites with /29 or higher always have the gateway at the top end of the range. Not much of a difference but it shows that not all are programed identical.

My concern is why a gateway is causing lag before even going over the WAN. I have fought with inconsistent connections at some of our sites for about 6 months and now that I have found this odd lag in some of the gateways it all correlate with the sites with performance issues. We notice most in VoIP calls. They are very susceptible to jitter and the gateways are introducing massive jitter on top of any they may naturally occur routing over the internet. It is also noticed with packet drop when simply surfing the net. A site like Amazon for example will often load the text and then stop as packet loss is to a point the connection was given up on. As you can see here http://s21.postimg.org/bb9kf58jr/Untitled_2.jpg. Before someone notices and jumps to conclusions, yes this screenshot was from my laptop connected wirelessly to our 802.11AC network with full signal but this occurs with hardwired computers as well and has nothing to do with the wireless network. The Wi-Fi on our DPC3939B is off and we use an enterprise class wireless system. All tests where done from hardwired computers to rule out Wi-Fi.

I have performed one more test to rule out wiring or routing. I have connected a laptop to one of the ports on the back of the gateway and the other port to our router. Ping gateway and you get a high ping. Ping laptop which must travel through the same router and over the same cable plus another cable to get from the gateway to the laptop and pings are sub 1ms. Latency must be in the gateway itself. Both Gateway and laptop are on 172.16.0.0/24 network while on the other side of the router we are all 10.10.0.0/24. If there was a problem with the router or cabling it would have exhibited the same high pings to the laptop. It actually should have been worse as it must travel the same path plus an extra cable.

I have a new SMC gateway on its way as we have had the best luck with them. I will be seeing how willing the tech will be to me testing some stuff before and after the statics are loaded. I am really hoping to see solid 1ms ping times till the statics are loaded then high ping times so I can narrow it down to a configuration problem. I wish I could see what they are actually sending the gateway so I could make this determination a little more specific. If it is a programming error it could help a lot of people who may not even know they have a problem as they are just used to it. If it doesn’t change between static programing and non I am out of Ideas.

If anyone has any other tests they think I should perform before and after static programing let me know and I will see how much time the tech is willing to let me have.

Today I had my DPC3939B replace with an SMC (my gateway of choice on Comcast). 

The Tech was awesome. Noting that he himself didn’t have an exceptional knowledge of how things were done but that he had responded many times to the same complaint of half loading web pages and laggy internet. The solution was always the same, ditching the netgear or Cisco in favor of the SMC. 

I was impressed that he not only was going to let me run all my tests but wanted to know what and why I was doing them to help him in the future. His healthy curiosity is exactly what we look for in a network engineer. Book smarts seems to work against you in the networking world. Rarely do book smarts translate into the complex world of networking well. Is more about quick adaptability, a grasp on concepts and a bit of common sense

The problem is that all tests were inconclusive.

Gateway before any programing 0-1ms

Gateway after initial programing but no statics 0-1ms

Gateway with statics assigned 0-1ms

Gateway with DMZ to router 0-1ms

So it performed perfectly but I was really hoping for problems after static load so I could put my finger on the problem and maybe get someone to look further into this. In retrospect I wish I would have rerecorded the serial numbers so maybe with some help we could have narrowed it down to a production run of defective devices.

Long story short its fixed. I didn’t get to determine if it was hardware or software but I lean more toward hardware since the same static assignment was sent to this gateway. My ping time over the WAN is excellent again. I am in Spokane Wa, and my connection to the SIX (Seattle internet exchange) is only 21ms. That is great considering I have 2 bonded 10gb dedicated fiber connections to the SIX downtown and my ping is 11ms. 20gb fiber vs my 100/20mb Comcast I would say 21ms is plenty respectable.

I guess the bottom line is if you internet doesn’t feel snappy, ping your gateway. If it doesn’t respond in less than 3ms have it replaced. 

One caveat to this is wireless connections. Do this test with no router in the way, directly connected with Ethernet to the gateway. That way no other lag is introduced.

I will continue to do testing at my sites with lag and have equipment replace as necessary. I will keep logs of all my tests if anyone at Comcast or elsewhere is interested.

Exact same issue which just started after two weeks of no problems and oddly enough same time as outage in my area. Ping any server on my LAN and get something like rtt min/avg/max/mdev = 0.046/0.054/0.061/0.009 ms; Ping the gateway and I get unacceptable rtt min/avg/max/mdev = 2.841/5.254/9.460/2.061 ms ; many tests much worse; first I though one of my cat 7 cables or one of my switches failed but that last test was a server connected directly to the DCP3939B gateway. This is a brand new unit manufactured 3/15.

When I was talking to tech support several hours ago they told me my modem was in bridge mode although it clearly isn't based on the modem's web page but I am also unable to place modem in bridge mode. Modem resets ineffective. Streaming any media impossible except for local and speed test show me around 80 mb down now  with the ugliest and dramatic saw tooth wave form.(pay for 100mb down and was getting around 7% faster) Up speed uneffected. I cant keep connection live from from remote locations; too many timed outs. 

Off topic but I was curious to see what battery was in this thing since the install tech stated I would get free replacements for life and Comcast would know when the battery was failing. Apparently it takes them awhile to do this since there was no battery installed in unit. But this also came from same Tech who said if there is a problem they have to send someone out within four hours and earliest I can get a Tech out is