[OpenVPN home] [Date Prev] [Date Index] [Date Next]
[OpenVPN mailing lists] [Thread Prev] [Thread Index] [Thread Next]
Web openvpn.net

Re: [Openvpn-users] performance numbers part 2

  • Subject: Re: [Openvpn-users] performance numbers part 2
  • From: "KT" <krt2941@xxxxxxxxxxx>
  • Date: Thu, 13 Dec 2007 21:49:00 -0800

> On Dec 14, 2007 9:44 AM, kt o <krt2941@xxxxxxxxxxx> wrote:

>> Client -> Server: (i.e. client runs netperf and server runs netserver)
>> Dev Tun, plaintext, no-LZO: 48.82, 49.62, 48.30, 49.70
>> Server->Client: (i.e. client runs netserver and server run netperf)
>> Dev Tun, plaintext, no-LZO: 63.70, 64.03, 61.06, 63.73
> This is the key difference I believe. For some reason, with this
> config, your openVPN setup is much faster sending data to the client.
> One way to dig up a little more information. Can you swap the client
> and server? Make the Opteron the VPN server, and the Athlon the VPN
> client? Then repeat this test. That way we could isolate if this is a
> hw/network card issue. My suspicion is that this has something to do
> with the number of interrupts generated - perhaps the client is faster
> at interrupt handling? We would also need to figure out the effect of
> XP and 2003.

That's a good idea... here's the numbers:

Opteron (Server) -> Athlon (client)
Dev tun, plaintext, no-LZO: 51.02, 50.57, 52.71, 51.93
Hamachi: 65.45, 65.92, 65.98, 65.50
No VPN: 91.15, 91.61

Athlon(Client) -> Operton(Server)
Dev tun, plaintext, no-LZO: 67.65, 67.22, 67.29, 67.32
Hamachi: 54.19, 57.29, 55.90, 55.64
No VPN: 89.39, 92.12

Strange that the absolute throughput values went up for OpenVPN.  The 
Hamachi and no-vpn absolute numbers barely changed.  But still, the same 
strange relative results depending on direction of traffic.

I'll see if I can borrow a co-workers WinXP desktop tomorrow - he's got a 
nice Core2Duo.  Then we can strip out the effect of XP vs 2003.


Openvpn-users mailing list