dslreports logo
site
 
    All Forums Hot Topics Gallery
spc

spacer




how-to block ads


Search Topic:
uniqs
8
share rss forum feed

bps

join:2013-02-11
reply to nickphx

Re: [AZ] Huge latency and packet loss issues in the Valley Area

I'm getting ~60% packet loss through 68.1.5.139. A sample mtr output to "google.com" is available here: »dpaste.com/hold/920123/


nickphx

join:2009-10-29
Phoenix, AZ

1 recommendation

it's been posted several times, cox's routers will de-prioritize or even drop icmp. If you're not seeing packet loss after that hop, there is nothing wrong.



NormanS
I gave her time to steal my mind away
Premium,MVM
join:2001-02-14
San Jose, CA
kudos:11
Reviews:
·SONIC.NET
reply to bps

I concur with nickphx See Profile. The truth is, every diagnostic trace packet is dropped by every router on a route. When they do return a response, it is a different packet from the one you sent; which requires extra processing time for the router. So they don't always take the time to respond.
--
Norman
~Oh Lord, why have you come
~To Konnyu, with the Lion and the Drum


Rakeesh

join:2011-10-30
Mesa, AZ
Reviews:
·Sprint Mobile Br..
·Cox HSI

Traceroute is just a standard ICMP echo (aka ping command) only we start the TTL in the IP header at just 1, and then increment it each time to find successive hops.

For those who don't know, the TTL is a value that decreased by one each time it passes through a router (aka a hop). When the value reaches zero, the router where it hit zero will send a return packet indicating that the TTL expired in transit, which includes its own IP address. This is done to mitigate downtime caused by routing loops. A side effect is that it allows traceroute to identify all of the hops between you and the destination (otherwise routers are invisible.)

Many routers are configured to ignore ICMP echo requests however, so when one lands at their interface with an expired TTL they just discard the packet and don't return a notice that it was discarded. This is typically done for security reasons rather than performance reasons. If your destination host also ignores ICMP echo, (All versions of Windows Server that I have used are configured to do this by default, by the way) you'll never know exactly how many hops it takes to get to your destination, so traceroute will just keep counting up until it reaches the max number of hops you specify (windows defaults to 30 IIRC).