[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Simple Security - Layered Filtering should be in the document



On 30 jul 2009, at 1:27, Gregory M. Lebovitz wrote:

if tunneling is the default avenue for building new protocols, then you can bet the under ground economy attack authors are going to be forming exploits on them. Which is why customers need tools that are able to block them

Surprisingly, that doesn't necessarily follow.

What's needed is a mechanism to protect hosts from packets that:

1. They have no interest in receiving
2. Could harm them

Note that this is a weaker proposition than in a corporate environment, where the local interest in receiving packets is not a relevant factor. However, in a home environment we must allow users to be stupid if they so choose.

As far as I know, there are no existing <something>-in-IPv6 encapsulations that mainstream OSes decapsulate by default. So any encapsulated packets will be dropped on the floor by a typical Windows or MacOS box, and as such there is no need to spend cycles to filter them out.

if so desired

Of course going beyond what's in the draft is always a possibility, but that has to require an explicit action by the user, we need to have a standard level of permeability so the makers of peer-to-peer applications have a standardized way in.

I know there is a school of thinking that says incoming packets of any kind except ones explicitly desired by the host as shown by previous outgoing packets are unacceptable, but the consequence of that line of thought is that many classes of peer-to-peer applications become impossible. In this situation, only a simultaneous open can work, and these require a control channel and are notoriously hard to get right.

However, I would love to apply such a policy to my mailboxes, both the electronic and dead tree types.