|
to iansltx
Re: [TWC] Deciphering tracertsaid by iansltx:First-hop latency doesn't matter, as long as it disappears after the first hop.
Explain please. |
actions · 2012-Oct-3 4:16 pm · (locked) |
hobgoblinSortof Agoblin Premium Member join:2001-11-25 Orchard Park, NY |
Unless you are playing a game or something on the first hop a high ping time to that is not a reflection of your performance to other sites. That hop can be configured either not to respond to pings, or do everything else its trying to do before responding which of course will be reflected in the trace. The last set the OP posted were perfect to the ultimate destination. The early ones were very ugly with the time outs but again if you exclude the time outs the ping time to the ultimate destination looks ok.
Hob |
actions · 2012-Oct-3 8:10 pm · (locked) |
|
If you have compared the tracerts from the primetime to the morning tracerts, there is a big difference in ping times. The OP mentioned his download speed usually in the non prime time @25Mbps and settles to 10Mbps at night. He is loosing 15Mbps at night due to a congested TWC gateway. The traceroutes represents the problem. For the OP to have a better service, TWC needs to split the node he belongs. This has been a long time problem in Queens, it appears there are still problem in some areas. |
actions · 2012-Oct-3 9:06 pm · (locked) |
|
hobgoblinSortof Agoblin Premium Member join:2001-11-25 Orchard Park, NY |
Look at the time at the last hop and then tell me that again. The only thing that matters is the LAST hop.
Hob |
actions · 2012-Oct-3 9:13 pm · (locked) |
|
Wrong loosing 15Mbps because of a congested gateway is not acceptable. We are not talking about the last hop. Everyone can reach the last hop but the question is for how fast. |
actions · 2012-Oct-3 9:17 pm · (locked) |
hobgoblinSortof Agoblin Premium Member join:2001-11-25 Orchard Park, NY |
"Wrong loosing 15Mbps because of a congested gateway is not acceptable. We are not talking about the last hop. Everyone can reach the last hop but the question is for how fast."
The trace shows exactly how fast you can reach the last hop.
The packet loss is a problem in the first routes with a router, with the router out it looks a lot better.
Much more troubleshooting needs to be done before we can reach the conclusion you did.
Hob |
actions · 2012-Oct-3 9:21 pm · (locked) |
|
Without the router 1 2080 ms 1511 ms 1186 ms cpe-24-xxx-xxx-x.nyc.res.rr.com [24.xxx.xxx.x]
With the router 1 4 ms 1 ms 1 ms 192.168.1.1 2 3332 ms 3124 ms 2629 ms cpe-74-xx-xx-x.nyc.res.rr.com [74.xx.xx.x]
From a congestion point of view, they are just the same high ping times, that were taken at different times. His non prime traces is a lot better in single or ten ms digit. Compared to our node, that is not congested even at night and because of it we get the same speed consistently at anytime. Here's the gateway's ping right now. 2. hop: 68.xxx.xxx.x (cpe-68-xxx-xxx-x.NYC.res.rr.com): 20 ms |
actions · 2012-Oct-3 9:38 pm · (locked) |
hobgoblinSortof Agoblin Premium Member join:2001-11-25 Orchard Park, NY
1 recommendation |
Sigh.
Lets look at the last hop
With the router Night:
21 20 ms 22 ms 22 ms lga15s28-in-f3.1e100.net [74.125.226.195]
11 * 14 ms 15 ms 170.149.168.130
With the router Morning:
16 19 ms 21 ms 22 ms lga15s28-in-f5.1e100.net [74.125.226.197]
11 15 ms 15 ms 15 ms 170.149.168.130
Without the router:
10 18 ms 14 ms 15 ms 170.149.168.130
14 20 ms 21 ms 19 ms lga15s34-in-f5.1e100.net [173.194.43.5]
As any technically minded person will tell you the only hop that we care about is the last one. No one is interested in the first one other than you. As I said previously the packet loss with the router is a concern.
Hob |
actions · 2012-Oct-3 9:48 pm · (locked) |
|
said by hobgoblin:Sigh.
As any technically minded person will tell you the only hop that we care about is the last one. Wrong, a technically knowledgeable person will look at every hop, any congestion at each route will affect the latency. And that's why traceroute helps technical people to find router problems if any. |
actions · 2012-Oct-3 9:57 pm · (locked) |
hobgoblinSortof Agoblin Premium Member join:2001-11-25 Orchard Park, NY |
"Wrong, a technically knowledgeable person will look at every hop, any congestion at each route will affect the latency. "
Ok I am done with this conversation. This discussion has been done to death multiple times over the years. You must have missed them.
Hob |
actions · 2012-Oct-3 10:11 pm · (locked) |
|
Multiple times dead, wow that's your best technical answer. I agree, I must end this conversation. Don't forget though you need to help the OP. |
actions · 2012-Oct-3 10:17 pm · (locked) |
hobgoblinSortof Agoblin Premium Member join:2001-11-25 Orchard Park, NY |
"Multiple times dead, wow that's your best technicl answer." A technical person can always spell technical correctly. I would suggest the OP visits » Time Warner Cable TV/Voice and posts relevant information in that forum. Good edit! Hob |
actions · 2012-Oct-3 10:22 pm · (locked) |
|
said by hobgoblin:"A technical person can always spell technical correctly. Hob Nope you're too late. |
actions · 2012-Oct-3 10:25 pm · (locked) |
|
BeachieWhere is Shelly Miscavige? join:2001-07-12 Saint Petersburg, FL |
It's not too late to fix [sic] loosing |
actions · 2012-Oct-3 10:40 pm · (locked) |
|
to bluepoint
said by bluepoint:Wrong, a technically knowledgeable person will look at every hop, any congestion at each route will affect the latency. And that's why traceroute helps technical people to find router problems if any. A traceroute directs UDP or ICMP at EACH HOP separately, and gets an ICMP response from EACH HOP separately. Each measurement of latency is for the time which that individual hop responded to the UDP or ICMP from the traceroute. If I am getting low latency from the last hop, then that is all that matters. Latency at any particular hop that doesn't continue is latency to the CPU of that particular router hop. Modern day service provider routers have the management plane separated from the forwarding plane. A router responds to traceroutes from its management plane. If that management plane is busy, it will delay the ICMP response to the traceroute. All of the other traceroute packets pass through the router's forwarding plane, which is controlled by ASICs or other dedicated CPUs within the linecards of the router and are designed to forward packets quickly. Maybe now you'll understand why high latency from a specific hop does not affect the entire round trip if the same latency is not being added to each hop thereafter. Latency caused by congestion would show up at every hop in the traceroute, because the forwarding plane (the linecards) have the actual ports that would be congested. Adding the congestion latency to each traceroute packet that is being sent to each subsequent packet after the congested port. This traceroute shows no signs of congestion latency. |
actions · 2012-Oct-4 7:21 am · (locked) |
(Software) OPNsense Ubiquiti UniFi UAP-AC-PRO
|
to bluepoint
said by bluepoint:said by hobgoblin:Sigh.
As any technically minded person will tell you the only hop that we care about is the last one. Wrong, a technically knowledgeable person will look at every hop, any congestion at each route will affect the latency. And that's why traceroute helps technical people to find router problems if any. In this case there is no evidence of really anything wrong in any of the trace routes except for that fist one with all of the *'s in it. The "First" hop with high latency just mean the device is putting ICMP directed at it and only at it on the back burner while it deals with other traffic, hence why the rest of the hops have a normal latency, if the high latency continued then yes there really is a delay there. What ever the problem is its not showing in the traceroute. |
actions · 2012-Oct-4 2:40 am · (locked) |
golden eagleAquila chrysaetos Premium Member join:2002-08-06 On a cliff |
to skuv
Whoosh. Okay that being said what could be causing my consistent connection issues? TWC seems to be of little help. As it turns out that "24 hour device watch" is a 14 day watch and short of sending someone out to check my line when I experience the issue - which TW doesn't send techs out at that hour - I may be sol. They told me last night that my "modem is offline" but I was connected to the net. Certain pages did load it just took forever. I mean I understand that as demand grows during the day my speeds will be lower but starting last week and coming to a climax the last two nights nights I've been essentially regulated to an AOL dial-up connection. I couldn't even log in here last night to read thru this thread.
All I know as a novice is that my issue is consistent -every day and there's a pattern to it - in the evenings. I'd like to get more insight so I can understand what the cause is. |
actions · 2012-Oct-4 9:23 am · (locked) |