softether client vs microsoft l2tp

Post your questions about SoftEther VPN software here. Please answer questions if you can afford.
Post Reply
cjarrett
Posts: 3
Joined: Thu Jul 02, 2020 4:09 pm

softether client vs microsoft l2tp

Post by cjarrett » Thu Jul 02, 2020 4:21 pm

With more users working remotely lately performance issues have arisen with our Softether VPN implementation. I have the majority of users utilizing the softether client but have done testing with Windows builtin VPN client over L2TP. The L2TP connections have consistently better performance than the softether on port 443. Is there a configuration settings I'm missing somewhere as everything I've read said the softether client should provide the best performance.

The server is Server 2019 hosted on VMWare 6.7. The network cards are using vmxnet3. I've tried adjusting the number of connections on the client to the recommended 8 and other variations. I've also tried with and without half duplex enabled.

Any insight as to the issue would be greatly appreciated.

centeredki69
Posts: 305
Joined: Wed Sep 18, 2013 1:49 pm

Re: softether client vs microsoft l2tp

Post by centeredki69 » Tue Jul 14, 2020 11:25 pm

I had this issue with a SE server I had on a cloud hosted Linux VPS and I was using SecureNAT. I found I had to lower the MTU setting from 1500 to I believe it was 1260 in SecureNAT. It only happened with this particular VPS provider as I had many other windows and Linux servers running SE at other VPS host providers also using secureNAT. I never understood why.

darwin.ranzone
Posts: 6
Joined: Wed Jul 15, 2020 2:38 pm

Re: softether client vs microsoft l2tp

Post by darwin.ranzone » Wed Jul 15, 2020 3:17 pm

why would you use a windows 2019 for a softether vpn server?
get more with less in a centos box

cjarrett
Posts: 3
Joined: Thu Jul 02, 2020 4:09 pm

Re: softether client vs microsoft l2tp

Post by cjarrett » Wed Jul 15, 2020 3:23 pm

Per softether documentation.

“ When developing the SoftEther VPN Server, SoftEther VPN Project develops the programs as well as basic debugging and optimization on a Windows platform and then ports these tasks to other operating systems. That is why the performance of the Windows OS kernel scheduler and network protocol stack is equivalent to or slightly better than those of Linux, and not only matches the performance of UNIX operating systems, but exceeds it. Using the Windows version of the SoftEther VPN Server also enables VPN Server operation with the least limitations.
In addition, while specific software may not work properly for Linux and other UNIX operating systems depending on the kernel version and differences between the distribution and various library systems, Windows operating systems guarantees a certain degree of uniformity in terms of the operation of system APIs and user-mode libraries such that the SoftEther VPN Server can operate safely on both old and new versions of Windows.
Accordingly, SoftEther VPN Project recommends using the SoftEther VPN Server on a Windows platform where no other technical or cost issues are involved.”

https://www.softether.org/4-docs/1-manu ... quirements

cjarrett
Posts: 3
Joined: Thu Jul 02, 2020 4:09 pm

Re: softether client vs microsoft l2tp

Post by cjarrett » Mon Jul 20, 2020 7:37 pm

centeredki69 wrote:
Tue Jul 14, 2020 11:25 pm
I had this issue with a SE server I had on a cloud hosted Linux VPS and I was using SecureNAT. I found I had to lower the MTU setting from 1500 to I believe it was 1260 in SecureNAT. It only happened with this particular VPS provider as I had many other windows and Linux servers running SE at other VPS host providers also using secureNAT. I never understood why.
Do you recall how you determined the proper MTU size? using a ping test I come up with 1472 + 28 = 1500 so by that the 1500 should be correct?

centeredki69
Posts: 305
Joined: Wed Sep 18, 2013 1:49 pm

Re: softether client vs microsoft l2tp

Post by centeredki69 » Mon Jul 20, 2020 7:58 pm

Honestly I chose the number based on a post from some forum over a year ago. I tried 1400 something an I believe it was 1260 That did the trick. The server is no longer in operation so I can’t check. I didn’t use any formula. I also have no idea why I had to change it. You either will notice an obvious increase in performance. Or it is not what is causing your issue. Your issue seemed similar to mine so I thought I would mention it.

Post Reply