When working with Traffic Shaping, on the System Services > Traffic Shaping Settings page, it has a field to fill in "Total available WAN bandwidth". Is this for: a) upload, b) download, c) the sum of upload and download, or d) something else?
It appears that Traffic Shaping is much more useful on upload for a couple of reasons. First, the up bandwidth to the ISP is pretty much the bottleneck. LAN speeds are generally much faster so it's taking the higher-bandwidth LAN traffic and trying to stuff it into WAN bandwidth that is a problem. Second, the XG sees all of the up-bound traffic and can queue/drop intelligently, but on the down-bound traffic the best the XG can do is to drop traffic and if it's connection-oriented (TCP) the unacknowledged packets will slow down the far end. But that's only some traffic and only indirect and binary (Fast/slow) control.
So my assumption is the Total WAN bandwidth would be referring to A (upload bandwidth). But then again, you can set limits/guarantees on both directions so how could the XG make even semi-reasonable guarantees on download if it has no idea of download bandwidth.
Any tips on setting the "Total available" or on the effectiveness of traffic shaping down (i.e. from WAN) traffic?
The times I've had the worst problems are during video conferences, when someone initiates a big download from a fast site and it makes incoming video stutter a lot. If no videoconferencing is taking place, more power to the downloader. I'd really rather approach it as a guarantee for video rather than a limit on downloads. (Not to mention there are many download options to cover.) So is the overhead -- if any -- of traffic shaping actually more of a parasitic loss than is worth it for the once-a-month-ish big download hurting videoconferencing?
(I'm beginning to think that while traffic shaping has been fun to finally figure out, it may not actually provide benefits in my use case.)
This thread was automatically locked due to age.