Trust Erosion: Dealing with Unknown-Unknowns in Cloud Security

Abstract

Although todays average cloud computing environment may incorporate security in most aspects of its design and infrastructure, the mere operation of the network exposes it to attacks. A typical attack starts with probing for weaknesses and/or vulnerabilities that can be exploited. And it is at this stage that the battle seems to be already lost, as the average network is insufficiently equipped-mostly for economic reasons-to even know that they are under probing, let alone thwart an attack. In many cases, cloud systems are caught unaware of situations where friends turn into foes, nullifying established security measures. Threats will always dwell on new (previously unknown) methods to compromise established security measures (i.e., a rat race between defenders and attackers, particularly well-resourced attackers). These methods largely fall outside the adapted models used by current security measures that protect cloud-based systems. After-the-fact analysis has driven security researchers to extend models to include assumptions about newly discovered threat(s). Solutions are then designed to deter these new threats. These models may also be generalized with additional measures mapping futuristic predictions-these are also referred to as known-unknowns.