Allocating resources for cloud networks
Georgios Paschos, Huawei Research, Paris
12-1pm 19th Oct 2018
In the era of virtualization, network management becomes flexible and network services can be deployed as virtual networks by means of software. A recognized bottleneck in this process is the computation of good resource allocation strategies as the network environment evolves. In this talk I will explain a key mathematical problem called virtual network embedding, whose solution allows us to determine efficient instantaneous allocations of resources for virtual networks. After discussing its complexity, I will focus on a convex variation, where the goal is to scale the resources of the virtual networks in a fair manner. I will devote the last part of the talk to the problem of adjusting the allocations as time evolves. The talk aspires to make a connection between popular tools from optimization and machine learning, such as the ADMM algorithm and the framework of Online Convex Optimization (OCO), to the existing practice in traffic engineering.
Since Nov '14, Georgios is a principal researcher at Huawei Technologies, Paris, France, leading the Network Control and Resource Allocation team. Previously, he was at MIT, LIDS ('12-'14) and CERTH-ITI ('08-'14), and he also taught in the University of Thessaly, Dept. of ECE as an adjunct lecturer for the period '09-'11. In '07-'08 he was an ERCIM Postdoc Fellow in VTT, Finland. He received his diploma in ECE ('02) from Aristotle University of Thessaloniki, and his PhD degree in Wireless Networks ('06) from ECE dept. University of Patras, both in Greece. He serves as an associate editor for IEEE/ACM Trans. on Networking, IEEE Networking Letters, guest editor of JSAC-SI on Caching, and as a TPC member of INFOCOM, WiOPT, WCNC, and Netsoft. His research interests include convex optimization and machine learning theory with applications to networking paradigms.