Trust key to Internet security

Five rings of trust could be the key to a more secure, more useful Internet

A few of my previous columns discussed my vision of creating a more secure Internet. It involved replacing the Internet's default anonymity with pervasive authentication, from the hardware initialization, through the OS and all applications, the user, and ending with a verifiable network stream. It is my strong belief that without a complete overhaul of default authentication, malicious hacking is going to continue indefinitely.

universal trust rings
And as expected, several readers complained about the obvious invasion of privacy my plan would entail; it's a duality I wrestle with as well. But an even bigger obstacle is the need to upgrade and replace nearly every involved software and hardware component. The computer world's normal evolutionary growth patterns don't fit nicely into my revolutionary changes. Few can afford to throw away the old and buy all new, much less have that occur across the globe all at once. It pretty much guarantees that my idea will never see the light of day … at least not for another 10 to 20 years.

[ RogerGrimes's column is now a blog! Get the latest IT security news from the Security Adviser blog. ]

What we need is "rings of trust" for Internet communications. At the OS level, the concept of protection rings has been operating on "trust rings" for decades. Ring 0, called kernel or supervisor mode, is where the OS kernel and device driver code runs. Only code in Ring 0 can directly communicate with hardware. Most user programs run in the least trusted Ring 3 layer, called user mode. Protection rings are built into most modern-day CPUs and are a part of most operating systems.

Internet Explorer even extends the concept into the Web browsing experience. It has five different security zones (Local Computer, Trusted Sites, Intranet, Internet, and Restricted) with customizable security settings. A single Web site or IP address can belong to only one zone at a time. NAP/NAC (network access protection/network access control) products introduce a similar model. Every unverified node is shunted to a limited-access quarantine network until after successful authentication.

My idea extends the NAP/NAC binary model into more rings of trust. Essentially, any user and computer meeting the highest level of trust would be placed into the highest trust ring. This would require verified hardware, an irrevocable unique machine ID, a verified boot sequence, verified OS files, authenticated applications, two-factor or more user authentication, and unmistakable network traffic. Users and computers on the old (that is, current) model would be placed in the lowest level of trust.

Every connection point along the way contributes to the overall trust ranking given to an end-point node. If your edge router doesn't do egress filtering, your ranking -- no matter how trustworthy -- is diminished. Hop on the Internet using an ISP known for hosting more than its fair share of bots, and your ranking goes down. Use a network with verified source routing, anti-DDoS protection, and antispoofing mechanisms, and your ranking goes up. 

Each participating node will get an overall trust rating along with individual component ratings. For example, user trust ranking might be a five because three-factor authentication is used. Applications and executables are signed, trusted, and approved, giving the application level a ranking of five. However, the hardware isn't verified and the MAC address can be spoofed, giving the computer a hardware ranking of two. A relatively secure network path is used between source and destination, giving a network ranking of four. The various component rankings end up making an overall trust rating of four. The destination network node receives a copy of the originator's trust rating and transmits its own trust ranking to the originator. The destination node can decide to treat traffic differently based on its own trust rating schedule.

Let's suppose the destination host is an e-mail server. Any traffic it receives from an origination node in the highest trust ring is immediately forwarded onto the recipient with no inspection. E-mail arriving from lower rings of trust can be forwarded to other computers for higher levels of inspection (anti-spam and anti-malware, for example). This is similar to the concept used in SPF and SenderID frameworks to verify originating e-mail addresses, extended pervasively across the entire OSI model. Not surprisingly, end-user applications would have to undergo a whole new level of programming to participate in the highest trust rings.

In my vision, a consortium of the world's best security experts, with vendor participation, would develop the various security mechanisms and determine what mechanisms earn what layers of trust. Both the secure and the insecure Internet versions could coexist. Want faster service? Become more trustworthy. Want to stay on the unprotected Internet? Do continue, but don't expect the fast connection to nodes residing in the higher rings of trust.

Boy, am I brilliant.

No doubt my "unique" idea was thought of 40 years ago by hundreds of greater minds and is already the basis of dozens of other proposals and products. That I can accept. My questions are, Why we haven't done more with it. Why are we continuing to allow the malicious minds of the Internet to use our online world as their perpetual bank account? Why don't we care enough to demand more?