RE: [smartBridges] Bandwidth Throughput

[phantam]

Title: Message

Very plausible we had a customer that we were doing local loops for with our ATM network and they were seeing ok performance with there lil 40$ switches we switched these over to those nice 250$ 2400m hp switches and configured up some nice vlans and stuff and the network has begun performing a lot more professional but that was just in our case.   Chris   From: Peter Cartwright [mailto:[EMAIL PROTECTED] Sent: Wednesday, July 16, 2003 1:25 PM To: [EMAIL PROTECTED]   Tom,   Sorry, now turned off.   Hmmm... since I am using switches not routers (and cheapo switches at that!) I wonder if that might have something to do with it?  Could be the packets arrive at the switch and for whatever reason, it takes a while to work out what to do with them.  Sound plausible?  If it is that, I wonder if replacing the switch with a hub would make any difference?   Peter     Loop Scorpio Ltd providers of Ledbury Broadband and Highnam Broadband www.ledburybroadband.co.uk www.highnambroadband.co.uk -----Original Message----- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]On Behalf Of Tom Haynes Sent: 16 July 2003 10:25 To: [EMAIL PROTECTED] Subject: RE: [smartBridges] Bandwidth Throughput Peter, please turn off your 'Read recipt requested' for your emails to the list.   I had a similar loss per hop when I first started. It ended up being the routers I used. I used a few linksys BEFSR41 4 port cable/dsl routers at two towers. They worked fine in gateway mode, however I had a 200Kb drop in throughput at each tower. After changing them to router mode and adding a few static routes, I no longer have any lag even after 3 hops.   internet--router--AP- - - - - -AP---router---AP- - - - - -AP---router---AP- - - - - -Client   One other thing that gets a few people, double check your 'basic rates' and 'auto fall back' from a computer connected to the device instead of over the network. There has been problems reported in the past when someone changed the setting remotely (From Chris I think. Right?). Everything looked right from the NOC using simpleNMS but it was found that the clients computer using simpleMonitor showed it still set at 1MB. If all else fails, set the radios to a fixed 1 or 2MB with no fall back. Also make sure that the AP (the one acting as an AP) is set to recieve all rates (check mark in all boxes). -----Original Message----- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Peter Cartwright Sent: Wednesday, July 16, 2003 3:45 AM To: [EMAIL PROTECTED] Subject: RE: [smartBridges] Bandwidth Throughput Tom,   Thanks for the feedback.   Default gateway ok, primary ports ok, all units on same subnet, APt's both at least 12' above ground, all units mounted on poles...   To be honest, I don't think it's a ping thing since the pings I get are very good between the units (generally 1 to 2mS per hop).    Also, no sign of errors etc in the SB monitor stats.   The info I posted was the closest we have got yet to a structured test, though we have seen similar things in another area where we have units (small urban town) where the distances tend to be a few hundred metres using roof top aerials.  The effect seems constant across a variety of installations which is what has me puzzled.  Prior to taking up with SB we used DLink 900AP+ units at both ends and had noticed similar effects.   I find it all quite worrying since I had expected our bottleneck to be our 2Mbps backhaul pipe and not the 11Mbps 802.11b network!   Since we are in blighty, we tend to mount our aerials a few feet above roof height and not on towers (would be next to impossible to get permission and even if we did the money we would need to pay landowners etc would destroy the business model).   Putting science to one side, it "feels" like there is something happening when the packets hit the 802.11b interfaces such that throughput decreases significantly at each hop.  I had wondered if it might be hidden node effects, but switching on RTS/CTS (default is off) has not seemed to make any difference.   Peter     Loop Scorpio Ltd providers of Ledbury Broadband and Highnam Broadband www.ledburybroadband.co.uk www.highnambroadband.co.uk -----Original Message----- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]On Behalf Of Tom Haynes Sent: 16 July 2003 05:36 To: [EMAIL PROTECTED] Subject: RE: [smartBridges] Bandwidth Throughput Make sure that the default gateway correctly in both units (closest router to AP). Primary port on AP should be ethernet, on APcb should be wireless. Make sure you are using IP numbers to test with (eliminates DNS and NB delays). Make sure your souer is configured properly. make sure all units are on the same subnet. Use trace route or VisualPingPlot to identify where the lag is taking place. Make sure you "open field" has no interference or large metal objects. Keep your APt's at least 6' above ground to reduce ground interference. Make sure the units are mounted on a pole instead of sitting on a table. It's all in the settings, read up on what each one does and you should be able to get it to work. Good luck! -----Original Message----- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Peter Cartwright Sent: Tuesday, July 15, 2003 8:20 AM To: [EMAIL PROTECTED] Subject: [smartBridges] Bandwidth Throughput {this is my first try on this email-forum, so I hope it works!}   What sort of throughput do you all get with the Smartbridges kit?  We have been noticing some strange results and have been trying to pin down what is happening.   Closest we have got to a real test is as follows:   Internet connection: 2Mbps :: ::=== computer (1) (connected via an Ethernet switch) :: AirpointProTotal (set as access point) :: 600m across basically a level open field (good RSSI and Link Quality) :: AirpointProTotal (set as client bridge) :: computer (2)   We first ran a throughput test just over the link using QCheck.  This gave us around 1.2Mbps to 1.4Mbps from computer (1) to computer (2), but strangely over 4Mbps the other way.   Next test was using toast (www.toast.net/performance) to measure throughput onto the Internet proper.  Computer (1) would get around 1.2Mbps, but computer (2) could only manage 0.6Mbps.   In both tests, we would run first one computer then the other etc to ensure that the same general conditions were applying.   This seeems strange to us for two reasons:   (1) why should the radio link throughput be so different in the different directions?   (2) why should the Internet throughput half when a single Smartbridges link is added (which according to the first test is able to deliver throughputs of well over 1Mbps).   When we add in double/triple hops, the Internet throughput drops even further.  For example, adding an AirpointPro Outdoor - Airbridge Indoor link as a second hop reduced the toast test to only 350Kbps.  To me, this suggests that it is 802.11b that is doing something.   Anyone noticed anything similar, or have any ideas?   Peter     Loop Scorpio Ltd providers of Ledbury Broadband and Highnam Broadband www.ledburybroadband.co.uk www.highnambroadband.co.uk