The Internet has enormous resiliency against damage. Unfortunately, it has very little against malice -- and that's becoming an ever-larger issue for IT managers.
While it's not true that the Internet was designed to withstand damage from a nuclear war, it can handle external threats with its decentralized, packet-based architecture.
However, the people who designed the network trusted each other, and by extension, all the other people who would use the system. They were worried about communication, not security, and they didn't think in terms of intrusion from within. As a result, the Internet is frighteningly vulnerable to malicious disruption from the inside.
Map of the Arpanet, a precursor to the Internet, c. 1974.
As the Internet has moved from being a tool for a few academics to a central part of our culture, we have been correcting this, but not nearly fast enough. There are still far too many vulnerabilities to attacks by bad guys -- attacks that can do anything from making a Website unavailable to bringing the whole Net to its knees.
The process of fixing this is painfully slow, in part because it's hard to change something as big as sprawling as the Internet has become and partly because the vast majority of IT managers and administrators are blissfully unaware of just how vulnerable the whole jury-rigged contraption really is.
At this point, change isn't so much a technical problem. We know what we need to change, and in most cases we have several different schemes of change to choose from. The real problem is generating enough awareness to force the change. In large part, this is going to be a job for IT management.
Indeed, if IT managers and other people using the Internet do not wake up and demand change, some bad guy is going to collapse the whole thing, causing chaos.
One of the contributing problems is that many of these vulnerabilities involve Internet minutiae that most IT managers, never mind ordinary users, aren't familiar with. The problems take a lot of explaining, no matter how dangerous they are.
Take, for example, the Internet's creaky routing system, and most specifically the Border Gateway Protocol (BGP). BGP is the glue that holds together the sub-networks that make up the Internet. Its job is to keep correct routing data so packets of information from inside an area are correctly routed outside the area. Since the Internet has no central directory for all routing, this is obviously a vital service. Unfortunately, it is also a vulnerable one. Or, as one Internet engineer put it: "The dirty little secret is that the Internet is still a handshake deal."
Theoretical diagram of the Border Gateway Protocol (BGP).
BGP is also responsible for updating routing information sent to connecting ISPs. If by accident or malice the ISP shares incorrect information, the routing system can blithely propagate it to all the other ISPs it is connected to. They blindly forward it on, and very bad things happen very quickly.
This isn't theoretical. We have already seen instances where by design or sheer dumbness the routing protocols have failed spectacularly.
In one notable example, a good part of the world was cut off from access to YouTube in 2008 because the authorities in Pakistan ordered their national ISPs to block access to some "anti-Islamic" video on YouTube. The ISPs took the easy way out and changed the BGP routing to dump all YouTube requests generated in Pakistan.
The short form is the ISPs made a mistake and transmitted the incorrect routing information to a regional center in Hong Kong as well as Pakistan. From Hong Kong, the new -- and incorrect -- routing information propagated to routers all over the world, and pretty soon everyone was trying to use the wrong directions to get to YouTube. The protocols accepted the changes as entered and knocked out access.
Now one can argue that cutting off access to YouTube might even be a public service, but the danger is that this can be done to any Internet address anywhere in the world. In fact, it could be done to hundreds or even thousands of them. The result would be ugly in the extreme.
Like most Internet vulnerabilities, this isn't a technical problem. There are a number of active proposals for modifying the BGP to make it significantly more secure, and most of them would work. (You can find a more complete discussion of the problem and proposed fixes here.)
Fundamentally, the problem is getting the momentum to actually implement the solution Internet-wide.
This and similar security holes aren't going to get fixed until IT managers everywhere demand it. It will take a broad effort to fix these problems and that will require broad and deep support from IT professionals of all stripes.
— Rick Cook is a prolific technology writer and author of the Wizardry series of fantasy novels.