![]() |
| |||||||||||||||
JSAC Special Issue 2013 : IEEE Journal on Selected Areas in Communications - Special issue on Switching and Routing for Scalable and Energy-Efficient Datacentre Networks | |||||||||||||||
Link: http://www.jsac.ucsd.edu/Calls/switchingandroutingcfp.pdf | |||||||||||||||
| |||||||||||||||
Call For Papers | |||||||||||||||
CALL FOR PAPERS
IEEE Journal on Selected Areas in Communications SWITCHING AND ROUTING FOR SCALABLE AND ENERGY-EFFICIENT DATACENTER NETWORKS Datacenters are the Internet brains, where scalability and power consumption take their toll. Cost-effective computing and storage are provided in datacenters due to economy of scale. Their key challenge is scalability, as a datacenter may house thousands of servers interconnected through many switches and routers, and the physical infrastructure is shared by multiple tenants, each deploying many applications. Datacenters process large amounts of multimedia traffic requiring guaranteed bandwidths and delays. Also, search engines and map-reduce applications generate large amounts of intricate traffic, which requires short delays, is bursty and induces incast effects, as many replies to a single inquiry are sent simultaneously to the node that initiated the inquiry. In both cases, current congestion control mechanisms incur unacceptable delays due to the retransmission of lost packets. In addition, incasting causes synchronization of traffic sources when packets are lost at the destination. Therefore, traditional congestion control needs to be either revised, or replaced by a new paradigm which may require an active role of switches and routers. Server virtualization and the corresponding proliferation of virtual machines (VMs) per node have raised several core issues: network management, network virtualization, and workload optimization. Tens or even hundreds of VMs per server are bound to exceed the capabilities of commodity switches. Mobility of VMs is needed to accommodate dynamic user demands, maintenance and reliability, as well as efficient energy consumption. However, moving VMs is challenging since it involves routing of gigabytes of data, as well as of the VM network states (addresses, QoS policies and requirements, and so on). Traditional routing and bridging protocols involve flooding mechanisms for the topology discovery and updates, which might clog a datacenter when VMs move. Efficient search of large amounts of data, VM mobility and multi-tenancy in flat datacenter networks incur an increasing burden on the lookup and classification because they complicate the address aggregation. Customers of datacenter services often require stringent service level agreements (SLAs), which necessitate deterministic QoS, including bandwidth and delay guarantees. Protocols designed for wide area networks support mainly best-effort traffic, and fairness is implemented only at the individual nodes. In a datacenter, flows of different customers must be isolated through the entire network, not only at the individual nodes, by means of novel QoS aware switching and routing, as well as network design based on realistic traffic models. There are also opportunities for performance improvements in datacenters with respect to WANs: the network topology is more flexible, propagation delays are low, and proprietary protocols can be used. New issues and opportunities of datacenter networks demand novel solutions for switching and routing. Of particular interest is the following, non exclusive, list of topics: • Scalable addressing, address lookup and classification algorithms • Enablers of SLAs in datacenters: Congestion control, flow isolation, and QoS • Performance optimized datacenters (PODs): flattened topologies, load balancing, efficient incasting, datacenter transport protocols, adaptive routing • Reliability and fault tolerance in datacenters • Energy efficient switching and routing in datacenters • Virtual networking and software defined networking, distributed routers and caching Prospective authors should prepare their submissions in accordance with the submission rules specified in the “Information for Authors” section of the JSAC guidelines (www.jsac.ucsd.edu/Guidelines/info.html). Papers should be submitted through EDAS (http://www.edas.info) according to the following timetable: Manuscript submission January 1, 2013 First review May 1, 2013 Final decision July 1, 2013 Final to publisher: September 1, 2013 Publication First Quarter 2014 Guest Editors: • Prof.Dr. Aleksandra Smiljanić (Belgrade University, Serbia), aleksandra@etf.rs, • Prof.Dr. Mounir Hamdi (Hong Kong University of Science and Technology, China), hamdi@cse.ust.hk, • Prof.Dr. Jonathan Chao (Polytechnic Institute of NYU, USA), chao@poly.edu, • Prof.Dr. Eiji Oki (University of Electro-Communications in Tokyo, Japan), oki@ice.uec.ac.jp, • Dr. Cyriel Minkenberg (IBM Research, Switzerland), sil@zurich.ibm.com. |
|