Cloud Computing & Virtualization SIG: Cloud and the Great Software System & Tera-scale Deep Learning
Topic: Fault Tolerance for the Netflix API
Presenter: Ben Christensen, Software Engineer, Netflix API Platform
The Netflix API receives over a billion requests a day which translates into multiple billions of calls to underlying systems in the Netflix service-oriented-architecture. These requests come from more than 1,000 different devices ranging from gaming consoles like the PS3, XBox and Wii to set-top boxes, TVs and mobile devices such as Android and iOS.
This presentation will describe how the Netflix API supports those devices and achieves fault tolerance in a distributed architecture while depending on dozens of systems which can fail at any time.
Bio Ben Christensen is a software engineer on the Netflix API Platform team responsible for fault tolerance, performance, architecture and scale while enabling millions of customers to access the Netflix experience across more than 1,000 different device types. Specializing in Java since the 90s and through years of web and server-side development Ben has gained particular interest and skill in building maintainable, performant, high-volume, high-impact systems. Prior to Netflix, Ben was at Apple in the iTunes division making iOS apps and media available to the world.
Topic - University and Google.
Presenter - Quoc Viet Le, StanfordTera-scale Deep Learning
Abstract: Deep learning and unsupervised feature learning offer the potential to transform many domains such as vision, speech, and NLP. However, these methods have been fundamentally limited by our computational abilities, and typically applied to small-sized problems.
In this talk, I describe the key ideas that enabled scaling deep learning algorithms to train a very large model on a cluster of 16,000 CPU cores (2000 machines). This network has 1.15 billion parameters, which is more than 100x larger than the next largest network reported in the literature.
Such network, when applied at the huge scale, is able to learn abstract concepts in a much more general manner than previously demonstrated. Specifically, we find that by training on 10 million unlabeled images, the network produces features that are very selective for high-level concepts such as human faces and cats. Using these features, we also obtain significant leaps in recognition performance on several large-scale computer vision tasks.
Bio: Quoc Le is a PhD student at Stanford and software engineer at Google. At Stanford and Google, Quoc works on large scale brain simulation using unsupervised feature learning and deep learning. His recent work was widely distributed and discussed on various technology blogs and news sites. Quoc obtained his undergraduate degree at Australian National University, and was research visitors at National ICT Australia, Microsoft Research and Max Planck Institute of Biological Cybernetics. Quoc won the best paper award as ECML 2007.
James Urquhart, VP of Product Strategy, enStratus and blogger
Six Degrees of Integration: Cloud and the Great Software System
As cloud computing and the Internet continue to create new opportunities to deliver integrated applications and services across political and organizational boundaries, it is creating a world in which every software application or service is connected to every other software application or service by a finite degree of separation. This, in turn, creates a very large complex adaptive system, and has profound implications to both software design and IT operations. What are complex adaptive systems, and how does one prepare to survive--or even thrive--in them?
Bio: James Urquhart is vice president of product strategy for enStratus, the leading enterprise cloud management solution. Named one of the ten most influential people in cloud computing by both the MIT Technology Review and The Next Web, and contributing author to GigaOm’s cloud coverage (http://gigaom.com/author/jurquhart), Mr. Urquhart brings a deep understanding of these disruptive technologies and the business opportunities they afford
6:30 - 7:00 p.m. Registration / Networking / Refreshments / Pizza
7:00 - 8:20 p.m. Presentations
8:20 - 8:30 p.m. Discussion