Am Montag 24 Januar 2011, 17:16:07 schrieb David Young: > If the ping times do rise when utilization rises[1], you may be > suffering from bufferbloat. This is not simple. "ping" is just requesting any host to "answer" on a recieving ICMP packet (usualy). There is no guarantee that the host (or his IP/ICMP stack) sends you and answer instantly when recieving your ping packet. This means beside network based effects "ping RTTs" could be rise even if the "pinged" OS means it has i.e. "more urgent things" to do then answering your "ICMP requests". One more bad thing is that ICMP could be handled very different from the "target" OS then i.e. other IP protocols / packets. So ping is not a really reliable tool for measuring transport times / delays of packets from interface to interface nor router to router node - but is is reliable to find out if a host "responses". Especially in access networks (i.e. by satellite) we've seen hardly varying ping RTTs within several networks not declaring anything real about the network performance nor "real" RTTs on a link. So leave a little bit patient if you rely on ping. cheers, Niels. -- --- Niels Dettenbach --- Syndicat IT&Internet http://www.syndicat.com T.-Muentzer.-Str. 2, 37308 Heilbad Heiligenstadt - DE --- Kryptoinfo: PGP public key ID 651CA20D Fingerprint: 55E0 4DCD B04C 4A49 1586 88AE 54DC 4465 651C A20D https://syndicat.com/pub_key.asc ---
Attachment:
signature.asc
Description: This is a digitally signed message part.