Comcast Cable I use uTorrent (also subject to QoS filters to reduce it's priority) but this problem seems unrelated since I can power off the uTorrent machine and get the same results.
I appreciate the "Packet Delay Variation" and "Jitter" comments. Fodder for the search engines. I'll check out SmokePing. Thank you. GGK On 03/08/10 02:12 PM, Benjamin Scott wrote: > On Mon, Mar 8, 2010 at 10:48 AM, Greg<g...@kettmann.com> wrote: > >> The network setup is pretty standard. Broadband connection ... >> > What type of Internet connection and who is the provider? For > example, "Comcast cable", "FairPoint DSL", "Verizon FiOS", etc. > > >> It seems to be an outbound problem (I can hear fine always) >> since the other end can't hear me for just a few seconds. >> > Any other traffic happening on the network? In particular, any kind > of server (HTTP, SMTP, FTP, etc.) or peer-to-peer file sharing > (BitTorrent, eDonkey, etc.)? Sometimes, the problem will be due to > outbound traffic, with the asymmetric nature of most consumer Internet > feeds able to saturate the outgoing channel long before the incoming > channel fills. This can confuse some traffic shaping/prioritization > mechanisms, which often assume a symmetric feed. > > >> What's the best approach to isolating or identifying the details of this >> problem? One obvious solution is to log some pings for a day or two. >> > This is tricky because what you're looking for is not throughput, > and not even latency (Round Trip Time), but rather, Packet Delay > Variation, AKA jitter. That is, you don't care how long it takes a > packet to transit, but you need to characterize *variation* in how > long it takes a packet to transit. > > I've seen SmokePing recommended in this sort of scenario before, > although I've never gotten around to trying it out myself. > > You might also traceroute to whatever gateway ViaTalk is using. > Maybe there's an obvious jump in latency, which, while not conclusive, > might be a clue. > > >> Bandwidth is good (based on the web based measurement >> tools) ... >> > Those tools, while not entirely useless, can be very misleading or > inaccurate. I am generally very hesitant to draw conclusions from > them. That said, throughput is rarely the problem with a single VoIP > feed, and I don't suspect it here. You only need about 50 Kbps for > that. > > -- Ben > > _______________________________________________ > gnhlug-discuss mailing list > gnhlug-discuss@mail.gnhlug.org > http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/ > _______________________________________________ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/