[Nets-seminars] Today's talk 25/11/11

Richard G. Clegg richard at richardclegg.org
Fri Nov 25 12:03:42 GMT 2011


Today's talk will be at 16:00 in GS/302.  The speaker is Paolo Costa 
from Imperial.

Applications and Network in Data Centers: Friends or Foes?

ABSTRACT

Since the early days of networks, a basic principle has been that applications treat the network as a black box. An application injects a packet with a destination address and the network delivers the packet. This principle has served us well, and has enabled the Internet to scale to billions of devices using networks owned by competing companies and running applications developed by different parties. However, this approach might not be optimal for large-scale Internet data centers, such as those run by Amazon, Google, Microsoft and Facebook, in which all the components are controlled by a single entity.

In this talk, I will describe two examples where a richer interaction between applications and network is beneficial for both. First, I will briefly overview our recent research in the context of providing applications with predictable performance in multi-tenant data centers. Then, I will describe more extensively our recent project, CamCube, in which, we have been looking at a different approach to build data centers, borrowing ideas from the fields of high performance parallel computing, distributed systems and networking. We use a direct-connect topology, similar to those used in HPC, and a novel networking stack, which supports a key-based routing functionality. By providing applications with a more fine-grained control on network resources, CamCube enables increasing performance and reducing development complexity and cluster costs.

During the talk, I will provide an overview of the CamCube platform and motivate its peculiar design choices. I will also describe the design and the evaluation of a number of services that we implemented on CamCube. These include a MapReduce service that provides significant higher performance than existing solutions running on traditional clusters.


-- 
Richard G. Clegg,
Dept of Elec. Eng.,
University College London
http://www.richardclegg.org/




More information about the Nets-seminars mailing list