Alex Shpiner (EE, Technion)
Wednesday, 20.11.2013, 11:30
Data center networks form one of the most challenging network environments for routers, due to their fast link rates, short propagation times, and high performance demands. In this talk, we analyze two router functions related to packet processing: address resolution and order preservation.
The first part of the talk deals with VM (virtual machine) address resolution in data centers. Data centers can run multiple VMs, and potentially place them on any of the servers. Therefore, a VM address resolution method that determines the server location of any VM needs to be provided for inter-VM communication. Unfortunately, existing methods suffer from a scalability bottleneck in the network load of the address resolution messages and/or in the size of the address resolution tables. We propose Smart Address Learning (SAL), a novel approach that selectively learns VM addresses for specific tenants, and solves this scalability bottleneck.
In the second part of the talk, we introduce novel scalable scheduling algorithms for preserving flow order in parallel multi-core network processors. The development of new processing features in advanced network processors has resulted in increasingly parallel architectures and increasingly heterogeneous packet processing times. This may lead to large packet reordering delays. Our suggested algorithms can significantly reduce reordering delay, while adapting to any load-balancing algorithm and keeping a low implementation complexity overhead.
Ph.D. Seminar. Advisor: Prof. Isaac Keslassy.