Throughput when bridged slower than NAT

I have a network that is bridged to clients. Mikrotik all the way through. CCR ↔ Sextant5 ↔ Toughswitch(at the POP) ↔ Sextant5 ↔ Sextant5 ↔ RB2011rm(all bridged ports) ↔ RB951-2n(CPE’s)

Bandwidth Test from the RB2011 to the CCR gets at lest 100mbps throughput in either direction. If I run a Bandwidth test from one of the CPE’s back to the CCR, the throughput is only to 30mbps down/40mbps up. I thought it was a horse power issue with the CPE but I get similar speeds when running speedtest.net on my laptop.

The problem is if the RB2011 acts as a DHCP server, throughput improves dramatically (100mbps+/100mbps+). Only when the client is connected to a bridged or switched port the throughput is slow… Any ideas? I’ve tried with ethernet auto negotiation both on and off. No firewall rules. No Mangle Rules. The 2011 is acting as a dumb switch. I’ve tried using gig ports as well as 100mbit ports. Additionally I’ve tried using Bridging and Switching. No significant change.

Heres a bandwidth test from the rb2011. Speeds are identical for clients when the RB2011 is acting as a DHCP server.
UDP test NAT.png
Here is a speedtest.net test from my computer when attached to a bridged/switched port on the RB2011
Speedtest RB2011 Bridged port.png
Here is a speedtest.net test from my computer when attached to a port providing DHCP on the RB2011
Speedtest NAT.png

When the RB2011 is bridged, could one of the other devices connected to the 2011 be acting as a DHCP server and your connection using that device as it’s default gateway instead of the CCR?

Have you tried disabling all interfaces other than the uplink and the one connected to your test router to see if it behaves the same?

Even some other device connected to the 2011 claiming to own a static default gateway IP which is supposed to belong to the CCR could be hijacking your connection. Double check that the ARP entry for the default gateway on your 951 has the right mac address for your CCR.

If it’s not a hijacked default gateway issue, I’m out of ideas.

lambert,

Thanks for the notes. Ill double check disabling all ports other than the uplink and my connected port. I do not believe its a separate DHCP server as I have rogue DHCP detection set on my core but I will investigate…