Re: [strongSwan] Establish connection with DynDNS peer

2010-01-11 Thread Gerd v. Egidy
Hi Peter, A is the strongSwan peer with a static and public IP address. B is a Bintec VPN25 router with a dynamic address published via DynDNS. [...] A tries to bring the tunnel up. However, A fails since it tries to connect to the OLD IP address. A ping from A to B shows that name

Re: [strongSwan] Establish connection with DynDNS peer

2010-01-11 Thread Daniel Mentz
Peter Daum wrote: B is a Bintec VPN25 router with a dynamic address published via DynDNS. A tries to bring the tunnel up. However, A fails since it tries to connect to the OLD IP address. A ping from A to B shows that name resolution works perfectly. So A seems to cache the old IP address

[strongSwan] Dead Peer Detection

2010-01-11 Thread vivek bairathi
Hi All, I have a query regarding dpd's:- 1. When does ikev2 stack start sending dpd's? 2. When does it know that its time to close the IPSEC SA or IKE SA? 3. Can you tell me where is the handling for closing the IPSEC SA or IKE SA in case of no response to the dpd's? Thanks in advance.

Re: [strongSwan] Establish connection with DynDNS peer

2010-01-11 Thread Gerd v. Egidy
Hi Daniel, Btw, can you recommend Bintec's VPN25 router? Does it support NAT-T (NAT traversal), DPD and certificate based authentication? We tried Bintecs and the results were mixed. At first everything seems to be ok, DPD and certificates work. I haven't tried it behind NAT but I think I saw

[strongSwan] Performance (revisited)

2010-01-11 Thread J.Witvliet
Hi all, Just re-reading the performace-thread from some-while ago. Are there any numbers available on what somebody might expect to see (given a certain hardware configuration) Obviously, if one put a 10Gb card into an old 386-system, one can not expect the line complete saturated with

Re: [strongSwan] Performance (revisited)

2010-01-11 Thread Martin Willi
Hi, What is needed (cpu) to get 10Mbps Not tested, but maybe a Pentium class processor? 100Mbps, Pentium 3/4? 1Gbps, A recent multi-core processor should be capable of doing 1Gbps, but requires parallel crypto patches, see [1]. 10Gbps Not without hardware acceleration. Maybe the

Re: [strongSwan] Performance (revisited)

2010-01-11 Thread Vincent JARDIN
Hi all, Even with the most efficient software and kernel implementation, you cannot exceed 1Gbps of small packets. Martin has sent a reminder of some benchmarks. I would advice that you check (sorry if it is commercial !, so please avoid the troll on the list) the 6WINDGate SDS architecture which