Question about ping times and trace routes.

[Eug]

Just wondering. My first hop to Sympatico (my DSL ISP) is on average about 50 ms. This seems to be a limitation of the PPPoE setup.

However, if I use the Windows ping command to the Globe (ie. ping www.globeandmail.com -t) it ALWAYS reads 10 ms total. How does that work? Does this mean that once the route is established the first hop ping time is no longer valid? Ie. the ping times for the hops are not cumulative?

Trace Routing... www.theglobeandmail.com [199.246.67.250] with 32 bytes of data:
01] 061 ms 030 ms 120 ms 64.229.64.1 [HSE-Toronto-ppp173586.sympatico.ca]
02] 020 ms 010 ms 010 ms 206.108.100.25 [dis17-toronto63-fe2-1-0.in.bellnexxia.net]
03] 010 ms 010 ms 010 ms 206.108.98.97 [core1-toronto63-pos11-8.in.bellnexxia.net]
04] 010 ms 020 ms 020 ms 206.108.107.142 [core1-toronto12-pos6-0.in.bellnexxia.net]
05] 010 ms 010 ms 010 ms 206.108.97.70 [dis1-toronto12-pos6-0.in.bellnexxia.net]
06] 010 ms 020 ms 020 ms 206.47.214.162 [No DNS Entry]
07] 010 ms 010 ms 010 ms 199.246.67.250 [www.theglobeandmail.com]

Trace Route statistics for: 199.246.67.250 Packets: Sent = 21 Received = 21 Loss = 0 [0.00%] RT Times: Min = 10ms Max = 120ms Avg = 21.00ms

I am using a Windows 2000 box with a Linksys 4-port DSL router and a cheap 10BaseT D-Link NIC.

In other words, if I'm already pinging servers at 10 ms does that mean cable is NOT going to speed up gaming? How does that first 50 ms factor into this?

 

[Russ]

I'm not sure exactly what you're asking, but response time (ping) is also a function of the server/s that you are connecting to. It's not just the speed of your connection, or the backbone along which the packets travel.

Even if all else is equal, a slow hop along the route can kill the response time.

Russ, NCNE

 

[Damaged]

Don't worry about the times to the first hop. It's simply that that device is slow to respond. If you can ping the final destination at 10ms everything is fine.

Traceroute computes times by increasing the TTL hop count to each device, not by pinging each device, then computes the time difference between the time it sent the packet and the time it received and ICMP host unreachable message. So, for the first device it sends it sets the TTL to 1, that device receives it, decrements, which will be 0, and means unreachable. Then it moves on to the next device, sending incrementing the hop count to 2, and so on and so on. This guarantees a final hop count of 0 and generates the ICMP message so that it can compute the time.

Notice how even though that one device is slow to respond that right after that all is fine? It's just that that device is slow to respond to those types of requests. Not unusual for a router busy doing other things. Sooo, don't sweat it.

 

[Eug]

OK thx. It would seem that really the ping command at the command prompt is really all that is important then.

 

[lumberg]

The tracert is used to show where in the network latency is occuring. If you get high ping times, run a tracert to see where the latency is occuring. Being an employee of an ISX(internet service exchange) we get calls and emails all the time about high latency slowing down things like games. Usually routers can be configured to speed up the connection by re routing data, or making configuration changes.

 

[Eug]

Yeah thx. I was just confused as to why my ping times seemed so great, but the pings while using tracert were so lousy for a couple of hops. I now understand that the tracert pings are not accurate for the reasons mentioned above. Ie. apparent lag is not real lag in many situations. Or am I out to lunch?

 

[spidey07]

Actually the trace route program on all windows OSs uses an ICMP-ECHO REQUEST. ie a PING.

1st frame has ttl of zero, 2nd frame has ttl of 1, 3rd frame has ttl of 2 just as damaged has explained. The time between the transmission of the ICMP-ECHO and the return of ICMP-TTL EXPIRED IN TRANSIT message is given in your tracert times. The three numbers are because tracert sends out three frames at a time to give you a slighly more accurate reading. Moral of the story is that busy routers have much more important things to do then send you an ICMP message, like routing frames. If you are pinging a server at 10 mils then consider yourself lucky with a fabulous network connection.

Also your ISP could be running caching servers. Get NEOTRACE from neoworx

Hope this helps.
spidey

 

[Damaged]

spidey07,

I should've mentioned that most routers deprioritzie ping/echo requests over actually routing packets (at least Cisco's do). Thanks for mentioning that; and for clearing up my explanantion. I just re-read my reply and realized it was pretty poor.

I had just woken up and had not even had a cup of coffee yet when I typed that reply. Bad idea to try explain something like that without the proper amount of caffeine.

 

[puritan]

You guys are both right. That is why traceroute does usually give close to accurate times. Becasue the router along the way are simply rejecting the packet because of it's TTL flag, not actually responding to the PING. Routers see ICMP echo requests, pings, as the lowest priority of traffic. Since they don't carry any real data in them. (So to speak) But that is true only to the router that is processing the ping, not the routers along the way. They are not looking that far up into the packet unless that are running some kind of QOS, Quality of Service, software that is checking and prioritizing the traffic.

Hope this all helps!