7100, slower than expected

AndyRH

What should I look at to fix the speed difference with my 7100? It is consistently slower than my old PC FW. About 100Mbs slower sending data.
ATT fiber 1Gb up/down
The 7100 just has a few rules, no packages. The test shown is using ix1 for the WAN, same results using eth1.
I ran many tests to make sure the results are correct.
The PC FW has a 4 port Intel 1Gb NIC.

JKnott

@andyrh

When I replaced the computer I was running pfsense on, I got the impression that for the same money, Netgear was on the slow side. I bought the computer described in my sig and that gave me a significant boost over the HP computer that I had been using . I have a 500/20 connection and was getting mid 500's down, but here's a recent speedtest with the new hardware. My ISP is generous with the bandwidth.

SteveITS

@andyrh If it isn't already, try plugging your PC into the 7100 directly instead of a switch.

AndyRH

I had previously tried direct connecting, eth2 has similar results. I also did the test using the same NIC on the PC on both FWs. The results did not change.

stephenw10

That's just using the default WAN and LAN config? Eth1as WAN?

AndyRH

Yes, with default settings the speed is the same as the graphs I posted. When I first noticed the slowness I did a factory reset and re-tested, no change. I also tried using the add-on port, igb3. I have since moved the WAN ix1, the host is directly connected to eth1, through my switch to ix0.

stephenw10

Hmm, OK. Check he output of top -aSH at the command line whilst testing. Make sure nothing it hitting 100% of one CPU core. It shouldn't be at 1G.

AndyRH

During the tests the 4 idle processes stayed in the top 4. The 5th process only got to about 36% (a little lower in the capture)

PID USERNAME PRI NICE SIZE RES STATE C TIME WCPU COMMAND
11 root 155 ki31 0B 64K CPU3 3 45.3H 98.57% [idle{idle: cpu3}]
11 root 155 ki31 0B 64K CPU1 1 45.3H 86.34% [idle{idle: cpu1}]
11 root 155 ki31 0B 64K CPU2 2 45.3H 78.14% [idle{idle: cpu2}]
11 root 155 ki31 0B 64K RUN 0 45.4H 68.10% [idle{idle: cpu0}]
0 root -76 - 0B 704K - 0 0:35 29.67% [kernel{if_io_tqg_0}]
0 root -76 - 0B 704K - 2 0:37 18.96% [kernel{if_io_tqg_2}]
0 root -76 - 0B 704K - 1 0:45 13.04% [kernel{if_io_tqg_1}]
8 root -16 - 0B 16K e6000s 2 163:10 5.31% [e6000sw tick kproc]

When I can I will make the network very quiet and make sure there is not an external thing I am not seeing.

stephenw10

Mmm, no problem there.

MTU mismatch, fragmentation issue perhaps?

AndyRH

Once I was able to take the primary FW offline and test on a quiet network the speed was what I expected. I am not sure why the new FW was not using the remaining bandwidth like I expected.
Perhaps the ATT gateway gives higher priority to the FW in the DMZ?

Sorry for the false alarm.

stephenw10

Mmm, interesting. Thanks for following up.