Identification and Disintegration of Single Points of Failure on the Internet
In this project, we investigate the Internet’s diversity and identify technologies that are too big to fail. This means that the respective technology is so widespread that its malfunction in response of an occasional error or a malicious attack significantly affects proper operation of the Internet or even a critical infrastructure. Accordingly, such events should be prevented best possible. The project’s insights allow to pinpoint technologies that have to undergo increased hardening efforts due to their importance or to be replaced by alternatives to increase diversity.
Wider Research Context
The Internet was originally invented as a distributed system maintained by many independent organizations in a decentralized way; this approach led to a highly resilient communication infrastructure despite using unreliable protocols. Today’s Internet however tends to be a homogeneous infrastructure as a high number of devices run the same hardware and/or software. Lacking diversity, these systems are prone to the same errors and attacks.
Our methodology consists of four steps:
- Investigating typical hardware and software configuration, we collect technologies that the Internet or connected hosts rely on.
- We infer assessment criteria for technologies that must not fail based on available attack and failure scenarios derived from real events as well as such described in the literature, and define parameters to be measured for the technologies under investigation.
- We perform Internet measurements investigating the Internet population to collect this data. From a technical perspective, we further develop active Internet measurement techniques, as well as passive approaches.
- Based on the gained results, we are able to determine technologies that are too big to fail and develop strategies towards a more diverse and resilient Internet overcoming the identified single point of failures.
This project is funded by the FFG.