Implications of Tightly Coupled Systems, Shared Infrastructure, and Unknown SPoFs
— In the past few years, the role of cloud computing, in which the infrastructure, platform, and even software used by IT operations are outsourced services, has become more prominent. The flexibility of cloud services certainly has its advantages, since they can be used when and only for how long they are needed, and be leased for prices charged in small increments of actual usage. However, these shared infrastructures also pose a risk that failures of a data center could cripple many services. This is supported by some analysts proclaiming that 2012 was the year of cloud (computing) outages. For many Internet services building on such cloud infrastructures, this creates risks that cannot be mitigated, as customers typically do not have much insight into the concrete building blocks of the used infrastructure and potential architectural SPoFs. This general issue, however, greatly extends this particular scenario of cloud computing. In recent
years, services have become increasingly coupled and integrated, which has also increased the vulnerability of Internet services due to common shared or cross-dependent infrastructures. Similar to the intensified linkages among actors in the financial market that led to the housing bubble burst in 2008, we might have created similar systemic or hyper risks [5] in Internet services which might explain the comparatively large magnitude of outages. The resilience of such tightly coupled systems is, however, in both general and specific for the case of computer networks and the Internet as its most prominent example still largely unknown. More research is
needed to understand risk and failure trajectories in these tightly coupled systems to develop
effective challenge mitigation strategies for Internet services operating under such circumstances