I have two sites, one with a 2820 and one with a 2830. They're both connected via WAN2 to ~10mbit EFM lines with an IPSec AH-SHA1 Auth VPN between the two.

Doing some speed tests with IPerf I'm seeing the speed through the VPN to be very very (~90%) much slower than the raw EFM to EFM speeds.

For example.

host behind router A (2820) via the vpn to a host behind router B (2830):

Code:

# iperf -c 192.168.30.1 -i 5 ------------------------------------------------------------ Client connecting to 192.168.30.1, TCP port 5001 TCP window size: 16.0 KByte (default) ------------------------------------------------------------ [ 3] local 192.168.11.1 port 35264 connected with 192.168.30.1 port 5001 [ 3] 0.0- 5.0 sec 728 KBytes 1.19 Mbits/sec [ 3] 5.0-10.0 sec 512 KBytes 839 Kbits/sec [ 3] 0.0-10.6 sec 1.22 MBytes 964 Kbits/sec

same systems but without the VPN (going direct to the external IP of the 2830):

Code:

# iperf -c -i 5 ------------------------------------------------------------ Client connecting to , TCP port 5001 TCP window size: 16.0 KByte (default) ------------------------------------------------------------ [ 3] local 192.168.11.1 port 49083 connected with 109.224.211.84 port 5001 [ 3] 0.0- 5.0 sec 5.44 MBytes 9.12 Mbits/sec [ 3] 5.0-10.0 sec 5.05 MBytes 8.48 Mbits/sec [ 3] 0.0-10.3 sec 10.5 MBytes 8.55 Mbits/sec

Wow! That's nearly 10 times faster!!! :o


System A has the default route set to the 2820, here's a traceroute across the vpn:

Code:

# mtr -r -c 1 192.168.30.1 HOST: systemA Loss% Snt Last Avg Best Wrst StDev 1. v2820 0.0% 1 0.2 0.2 0.2 0.2 0.0 2. 192.168.30.254 0.0% 1 35.8 35.8 35.8 35.8 0.0 3. systemB 0.0% 1 36.2 36.2 36.2 36.2 0.0

and direct to the external IP:

Code:

# mtr -r -c 1 HOST: systemA Loss% Snt Last Avg Best Wrst StDev 1. v2820 0.0% 1 0.2 0.2 0.2 0.2 0.0 2. no-reverse-dns-set.gradwell. 0.0% 1 21.9 21.9 21.9 21.9 0.0 3. no-reverse-dns-set.gradwell. 0.0% 1 36.6 36.6 36.6 36.6 0.0

Does anyone have any ideas? While I can see that using the VPN will have some overhead, dropping from 8Mbit to 800Kbit seems a little excessive!

(PS tests run the other way, from System B back to System A, show the same dropoff).

TIA