Before you decide to use CIPE, read:
http://www.mail-archive.com/cypherpunks … 09553.html
http://diswww.mit.edu/bloom-picayune/crypto/14258
http://diswww.mit.edu/bloom-picayune/crypto/14238
http://www.cs.auckland.ac.nz/~pgut001/p … ux_vpn.txt
http://www.netheaven.com/pmtu.html
Now, these are about version 1.5, and the current version seems to be 1.6, so some issues might be fixed. But you have to wonder, how safe is it to use a VPN protocol that hasn't seen much public (or private) research? Is it really worth the risk?
If you have some problem with IPsec, at least go and use OpenVPN instead of CIPE. Even though OpenVPN is also a non standard, not IETF tunnel mechanism, it at least uses TLS/SSL, which has seen a lot of public scrutiny like IPsec.
Another issue is the path mtu issue. You might run into lots of problems when to layers of IP start to fragment or drop packets, and ICMP's for path MTU discovery aren't reaching the right layer.
Olaf in his CIPE FAQ describes ciphers (AES, Blowfish, 3DES) and then says 'these are secure and so my protocol is secure'. That's a rather misleading statement. Especially because he is mixinger ciphers with algorithms and protocols. He then mentions the existence of DoS vulnerabilities, and other 'theoretical' vulnerabilities.
Frankly, IPsec and TLS/SSL are the two most widely used industry standards. They have been around for a long time, and are continiously shaken by the research community. I would not recommend to use CIPE protocol.
Paul
ps.diclaimer: I am an Openswan (IPsec) developer.