Page 1 of 1

Avoid bradcast to central node in L3 routing - high amount auf broadcasts

Posted: Fri Jun 14, 2019 8:49 am
by augur
Hi all,

I set up a L3 routing but a high amount of broadcasts tranferred to the central Softether instance.
14-06-_2019_10-42-10.png
Most of them seems to be UPNP lookups.

I changed the user credentials to avoid this in central instance but there are still broadcasts transferred. Any ideas about this?
14-06-_2019_10-49-15.png
Branch -----> Central (Switch Port 1) - Central (Switch Port 2) -----> internal network
Broacasts are in the segement underlined.

Regards,
Nils

Re: Avoid bradcast to central node in L3 routing - high amount auf broadcasts

Posted: Mon Jun 24, 2019 7:44 am
by cedar
Broadcast and multicast packets from the branch site network are sent to the central through the bridge.
To shut it off, please add another virtual hub to VPN server on branch site and add another L3 switch to it.

Re: Avoid bradcast to central node in L3 routing - high amount auf broadcasts

Posted: Mon Jul 01, 2019 8:39 pm
by Neil_P01
I've been looking at the Data Cost of keeping the VPN connection up, with all services off (Hubs at each end not connected to anything, switch and NAT off) I make it 15.5MB/Day or 450MB/Month. I have a switch at one site but will implement another switch at the second site and make sure DHCP traffic etc doesn't go across the link. In my experiments I tried different numbers of TCP/IP connections, this didn't make any difference. I see in the status GUI it said using 2 or 8 connections so I'm assuming it only uses the connections it needs (load dependent).

Re: Avoid bradcast to central node in L3 routing - high amount auf broadcasts

Posted: Mon Jul 08, 2019 8:44 pm
by Neil_P01
I've added a second switch so I now have one at each end of the site to site tunnel. Interestingly the data usage went up to to 70MB/day from 15.5MB/day.

I noticed it now says it us using all 8 TCP/IP connections whereas it wasn't before, even though it was set to 8. Most traffic is showing as broadcast packets however broadcast + unicast packets total only 7MB/day do I'm assuming the majority of the total traffic is just keeping the connections going.

Trying data compression next. If this doesn't help I'll change the number of TCP/IP connections to 4 and live with the data usage.

Does anyone know how SoftEther decides how many connections to use of the maximum?