Invitasjon til gjesteforelesninga: " Foundations of Cloud Computing Security " - Head of security research at Google, USA, dr. Úlfar Erlingsson
Head of security research at Google, USA, dr. Úlfar Erlingsson, vil holde en gjesteforelesning tirsdag 31. januar 2012 kl 12:15. Sted: Tabletten, Farmasibygget, UiT. Kart over Campus:
( http://www2.uit.no/ikbViewer/page/publikum/kart)
Alle interesserte er hjertelig velkomne.
Påmelding er ikke nødvendig. Dersom du har spørsmål om foredraget kan du kontakte professor Dag Johansen.
Vel møtt!
----------------------------------------------------------------------
Date: Tuesday January 31th 2012, 12:15
Place: Tabletten, Farmasibygget, UiT
Title: " Foundations of Cloud Computing Security "
Lecturers: Head of security research at Google, USA, dr. Úlfar Erlingsson
Abstract
Cloud computing offers a new approach to software and systems engineering with the intent to provide users with reliable, secure, and highly-available services.
Cloud services have an unusual, defining design characteristic: they are to provide continuous, ubiquitous access to rich functionality and ever-improving features, without any requirement for planned downtime. Most key properties of cloud computing result from this design characteristic: pervasive redundancy is required to handle independent failures, scalability and elasticity handles spikes in load, geographic replication handles user migration and local power outages, and so on.
Thus, the move to "The Cloud" does not merely constitute centralization or a change of scale. Instead, cloud computing is more correctly seen as a fundamental change in the nature of software and services―and this change creates both new risks and opportunities for security.
In particular, cloud software is uniquely well suited to security mechanisms based on specialization or artificial heterogeneity. Cloud services are designed so that new features can be rapidly added, without incurring system-wide unavailability due to changes to software or data. As a result, it must be possible for multiple different versions of cloud software to be concurrently executed, tested, deployed, and retired―even when each version uses different data representations and algorithms. Therefore, different cloud software instance can be quickly and easily tuned for different purposes, for example by adding specialized security enforcement or monitoring of some behavior aspects. And, for such security mechanisms, the redundant basis of cloud software can allow local, fail-stop enforcement to be a viable option, if coupled with appropriate denial-of-service countermeasures.
On the other hand, the security challenges of cloud computing are often greater than those of traditional software. Many cloud services are fundamentally based on untrusted, user-generated content from unreliable, unknown sources―and vulnerabilities in the processing of that data can be exploited to control individual service aspects, shared infrastructure, or even entire data centers. Cloud services often have only a virtual, indirect relationship with their users, and authentication and authorization of access to services and shared data remains very a difficult task. Even the origin and intent of cloud software itself is often uncertain, especially on cloud clients, where applications and extensions originate in open, moderated, or hand-curated software markets. Finally, above everything else, users' data must be protected―not only in terms of integrity and confidentiality, and giving users control, but also in terms of the privacy of each individual user―and this mandate must be fulfilled even for cloud services driven by aggregations of end-user data, such as search-query-completion suggestions, or speech recognition trained on user input.
Securing the cloud requires understanding and building upon the above fundamentals. However, this is not sufficient: security must be pervasive in the organizational structure and deeply integrated into the software development and operational processes of cloud service providers. Furthermore, every advantage the cloud offers―such as large-scale machine learning and data-driven abuse detection―must be utilized to help ensure security. A range of other techniques also help, including: formally-verified machine code, application-level interfaces designed to have strong, guaranteed semantics, and intrusion monitoring of software, networks, and other activity. Finally, cryptography must be used―properly, and comprehensively―at every single component and layer of the system, to establish everything from session validation to the integrity of auctions on advertisement exchanges.
Bio
Úlfar Erlingsson works on the computer security issues at the intersection of systems infrastructure, programming languages, and software engineering. He has often focused on low-level systems aspects, and on providing strong security guarantees―such as machine-code-level control-flow integrity―despite the presence of software bugs and vulnerabilities. Úlfar has done some of the earliest work on the security applications of software rewriting, operating-system-level virtualization and sandboxing, memory address-space randomization, transactional rollback, and server-side Web application policies. His work has directly impacted security mechanisms in the Microsoft Windows kernel and the Google Chrome Web browser, and indirectly impacted other, similar systems.
Úlfar currently heads security research at Google. Previously, and has previously been a researcher at Microsoft Research, an Associate Professor at Reykjavik University, Iceland, and led security technology at two startups: GreenBorder and deCODE Genetics. He holds a PhD in CS from Cornell University.
