What kind of throughput should you see with 3600 aIOS bridges?

[thefranmanatt]

I am seeing around 40mbps with the bridges showing m21 data rates with a clean 5GHz channel.  This is testing to a server on a 2960s (gig connections) to a 3600brdige  wireless over to a non-root 3600bridge  wired to an i3010 switch (fast E 100mbps) with the laptop client wired in (fast E 100mpbs). 

I'd think I could come close to the 100mbps port.  I didn't see much difference going from 20 to 40mhz wide on the channel.   Thoughts?

[Scott Fella]

Tough to say. When I have done testing I would first get a baseline of the throughput from laptop to server on a wired connection. Then in your case I would connect a laptop to the non root bridge and then see what throughput you are getting. At least you have a baseline to go off of. You might only see 50mb throughput when wired, you don't know unless you test. The only way to actually baseline throughput is directly connected to each bridge. Then you have an idea of the throughput you will have.

Sent from Cisco Technical Support iPhone App

[thefranmanatt]

It is lot faster client to client on the same switch bypassing the bridge.  I plan to bypass the 3010 and plug directly in to the bridge with the client tonight.  I was using it to power the AP but I have a power brick I am going to bring home for testing. 

I was hoping someone had used 3600 bridges like this before and had an idea of what good performance should be.  Are there any tweaks to frame or TCP windows sizes I should try to adjust for bridge use?  Could the 100meg port becausing issues?

thanks

[Scott Fella]

I have done a few indoor APs as bridges but only in close distances. Other than that, I've used actual bridges. It actually depends on what tool you use to test. I have used iperf to test between links.

Sent from Cisco Technical Support iPhone App

[thefranmanatt]

jperf (which is just iperf) and windows file share using the new windows 8 file share.  I'd think that 40mhz at m21 I should be able to hit 150mbps.  Of course the fastE port limiting it to 100ish mbps.

[Scott Fella]

Well what is the bridges negotiating at. Take that and divide that in half. That would be your max throughput from bridge to bridge testing. Then deduct you loss from the cable and the 100mbps port.

Sent from Cisco Technical Support iPhone App

[thefranmanatt]

200mbps when I hook the laptop up to the non-root bridge.  smoking!!!!!  Now to figure out why the i3010 is slowing me down so much.  In testing tonight I am seeing 69ish mbps through the i3010.

[Scott Fella]

What does jperf show you? Bandwidth Meter Pro is another good tool I use to gauge file transfer and FTP and or TFTP transfers.

Sent from Cisco Technical Support iPad App

[thefranmanatt]

It looks like these APs just don't like FastE ports.  Just for fun I put one of my LWAPs on 100/full (unrealed to the bridges).  It only runs about 40mbps on the file transfer.  I switch it back 1000/full and it goes back to 200mbps.

Anyone want to try the same test?

[Scott Fella]

Just remember that the AP's by default will auto negotiate.  If you hard set the switch duplex and speed, you can cause the AP to negotiate at 10/half/full.  I normally have no issues with the AP negotiating, but that is what I would verify when testing.

Thanks,

Scott

Help out other by using the rating system and marking answered questions as "Answered"

[thefranmanatt]

I couldn't see a way to set the hard code the AP from the GUI.  What I did was set the speed only to 100 on the switch.  I know this is a hack but I was able to get it connect at 100/full on both side.  I confirmed no errors on the port after the test.

BTW if one side is 100/full and the other is auto/auto you will normally get 100/half.  IIRC it can detect the speed but not the duplex.

good point though, thanks

[Scott Fella]

IIRC if you hard set a 10/100 port to 100/full the other devices usually falls back to 10/half.

On the bridge issue a show IP interface brief or a show interface

Sent from Cisco Technical Support iPhone App

[thefranmanatt]

http://www.cisco.com/en/US/products/hw/switches/ps708/products_tech_note09186a00800a7af0.shtml

[Scott Fella]

Well just verify what the AP negotiates to.

Sent from Cisco Technical Support iPhone App

[thefranmanatt]

I did confirm it in the GUI under the the AP's gig interface.  Again the port on the switch was error free.  If speed/duplex was wrong I'd see a lot of errors.

So it seems any time I have the AP connected to a 100/full port I get bad performance(70mbps) and if I am running through the port on CAPWAP it gets worse, (40mbps).

I think I'll just open a ticket.  LOL.