Section: Overall Objectives

Challenges

In this context two main challenges stand out:

• Scalability: As mentioned above, the Internet ecosystem is continuously expanding in both size and heterogeneity. Scalability was already a challenge in the last decade but solutions mainly focused on scaling one dimension at a time, e.g. increasing the capacity of network links or that of compute resources in order to face peak demand, even if it is infrequent. Such over-provisioning however wastes significant resources and cannot cope with future demand at a reasonable cost. Several experts warn about major Internet blackouts in the coming years [36], [33]. Scalability must be ensured across multiple dimensions to face of order of magnitude: more users, devices, contents and applications.

• Security: Security has gained a lot of popularity in the last few years because the Internet has become a lucrative playground for attackers with large numbers of potential victims and numerous ways to reach them. Advanced Persistent Threats (APT) [38] are the most sophisticated representatives of this evolution. Such targeted attacks do not rely on generic scenarios, usually described as a set of signatures. They are complex by nature and their investigation requires the analysis of various sources of data. At the same time, the generalization of encryption renders all deep packet inspection techniques obsolete and threat hunting becomes an even bigger challenge.

  Additionally, an underground economy has been developed by cyber-criminals. Finally, because many applications are now provided as cloud-based services, physical isolation is also harder with potential attackers able to act directly in the field.

The highly dynamic nature of the Internet ecosystem, the requirement for higher and higher scalability, and the rising security threats have shown the limitations of traditional approaches to address these challenges. Resist focuses on two complementary paradigms for achieving security and scalability:

• Elasticity refers to the ability of a system to scale up and down on demand. Elasticity of compute resources became more accessible with the advent of cloud computing. It has been recently leveraged in support of Network Function Virtualization (NFV) coupled with Software-Defined Networking (SDN). Understanding the dynamics of networked systems is critical in order to benefit from and efficiently orchestrate elasticity at all levels of the network, the system and the applications. On the one hand, elasticity facilitates scalability, as well as security by instantiating virtualized network security functions (e.g., firewall, IDS, DPI, etc.) on demand. On the other hand, it could increase the attack surface, which must also be addressed. This dilemma must be addressed. Moreover, issues inherent to elasticity such as the dynamic deployment and migration of resources bring new challenges in NFV environments since network functions are different from those of common cloud applications deployed in virtual machines and containers, e.g. in terms of network throughput.

• Resilience refers to the ability of a system to adapt itself when facing challenging situations. It is reasonable to assume that any system may face an attack for which protection mechanisms may fail. A comprehensive approach to resilience that considers not only the network and system resources but also the supported users and applications brings both benefits and challenges since users and applications can be very diverse, ephemeral and mobile. Applications are also deployed in dynamic environments like cloud platforms and are frequently reconfigured.

Resist ambitions to make large-scale networked systems more secure and more resilient, leveraging resource elasticity and assuming a highly dynamic environment.