To: PJ Strifas who wrote (30283 ) 2/13/2000 1:03:00 PM From: Scott C. Lemon Respond to of 42771
Hello PJ, > So basically we're talking about "self-regulation" in terms of > packet addressing to remove the ability of spoofed packets from > traversing from one network to another. This sounds like it could > become a very "do-able" solution. I wonder which alliance (IETF, > W3C, ISO, ???) would be able to bring all the participants > together.... So what we are really doing is bring the principals of the "physical world" to the Internet. If you think about how our society has evolved, we are now a point that if a country (or a community) does things which are unacceptable to the rest of the countries, then we issue sanctions against that country and cut off trade. In the Internet world, if an ISP is found to be a source of spoofed packets, or refuses to turn over a customer who has committed a crime, they will find themselves isolated from the rest of the Internet by sanctions. (I wanted to point out that we can't say they will be prosecuted, since they might exist in another country which is more tolerant of such behavior ...) And so "communities" will be born which allow/disallow various types of behavior, and the communications between them will be based on these behaviors and the netMorals and netValues that each have. ;-) > Still, is there nothing that can be done in terms of the IP stack > to add "checks/balances" so that tampered packets can be identified > and then dropped? There are a number of products out there which start to try and address this area ... I've been doing a bunch of research on this lately ... I believe that all machines will soon be running more sophisticated "personal firewall" software before long ... > Not to force this into a NETWARE discussion but in a NetWare > environment (using the Novell Client to log into a Novel network) a > user's session has unique identifiers that are used to create and > check packets exchanged between a workstation and the network. In > this way, packets can be verified in terms of whether or not they > have been modified (tampered with) during transmission. Yes ... there are some "rules" that can be adhered to to help verify valid traffic ... use TCP, packet signitures, etc. > Could it be possible that this type of functionality can be worked > into the current TCP/IP protocol? There are there already and available for use ... most of the problems start to come about with UDP ... so avoidance of this would stop a lot of the problems. And then improvement of the network configuration to help track offenders ... > Can we add additional layers such as (PKI or) digital certificates > to authenticate online sessions at the network, session and/or > transport level? (a way to create packet checksums as we have in > NetWare?) The best thing that I could see Novell doing would be to get the client connection to use SSL ... this would be a big step that is fairly simple to do. (And fix the client installation! ;-) The problem is that this has nothing to do with hackers that decide to flood a network with UDP packets ... these are the "connectionless" packets that don't require two way communications. As long as these can be sent, and spoofed, and ISPs do not implement "egress" filtering ... then we are going to have troubles. Again, relating it to the real "physical" world ... what you are suggesting is "Can't the government just stop the flow of drugs, illegal immigration, and terrorism?" ... I mean heck, they are spending enough money trying ... what's the deal? ;-) > I am not as fully versed in this topic as I would like to be > therefore it may hamper my understanding of what is possible and > what is not in the IP world. It's a cool subject area ... and I think that Novell *could* offer a lot in this space ... The problems that arise are due to the "fault-tolerant" nature, and "free form" of the Internet. There are going to be more and more "border checks" that will be done between ISPs, etc. in the future ... IMHO. > I surely understand this problem can be "attacked" from more than > one angle - perhaps that will be what enables us to find an > acceptable solution that balances performance and security. The biggest success, IMHO, will be in identifying sources of "bad" traffic and sanctioning them ... just like in the real world ... Scott C. Lemon