Flexible Multi-step Resource Relocation for Virtual Network Functions

Virtual machine (VM) migration is a technology dealing with the process of relocating a VM from one host to another. The relocation of VMs is usually assumed to be done in a single-step procedure, which however leads to a restriction on the degree of freedom. We therefore aim to consider a situation in which multi-step VM migrations are allowed. In this paper, by applying the concept of time-expanded networks [1], a novel optimisation model for virtual network function migration in multiple steps is introduced. Our model is based on the problem of virtual network embedding (VNE), which pertains to the mapping of virtual networks onto a capacitated substrate network and tackles the transition from one mapping to another in multiple passes. This publication is an extension of our previous work [2], which introduced the concept of soft bottlenecks and assumed deadline-restricted VM migrations. The work presented in this paper removes this restriction by allowing the migration time to vary in compliance with a variable VM image transfer rate, making the transition even more flexible. The obtained performance evaluation results underline the advantages of our multi-step VM migration model compared to any conventional single-step VM migration approaches.

[1]  Ramin Yahyapour,et al.  Cloud computing networking: challenges and opportunities for innovations , 2013, IEEE Communications Magazine.

[2]  Umesh Deshpande,et al.  Post-copy live migration of virtual machines , 2009, OPSR.

[3]  Kresimir Mihic,et al.  A system for online power prediction in virtualized environments using gaussian mixture models , 2010, Design Automation Conference.

[4]  Gustavo A. A. Santana Data Center Virtualization Fundamentals: Understanding Techniques and Designs for Highly Efficient Data Centers with Cisco Nexus, UCS, MDS, and Beyond , 2013 .

[5]  Flavio Esposito,et al.  GeoMig: Online Multiple VM Live Migration , 2016, 2016 IEEE International Conference on Cloud Engineering Workshop (IC2EW).

[6]  Prashant J. Shenoy,et al.  CloudNet: dynamic pooling of cloud resources by live WAN migration of virtual machines , 2011, VEE.

[7]  Thomas Bauschert,et al.  Robust embedding of VNF/service chains with delay bounds , 2016, 2016 IEEE Conference on Network Function Virtualization and Software Defined Networks (NFV-SDN).

[8]  Thomas Bauschert,et al.  A multi-step model for migration and resource reallocation in virtualized network infrastructures , 2017, 2017 IEEE Conference on Computer Communications Workshops (INFOCOM WKSHPS).

[9]  D. Kirschen Demand-side view of electricity markets , 2003 .

[10]  Philip J. Morrow,et al.  A Practical Evaluation in Openstack Live Migration of VMs Using 10Gb/s Interfaces , 2016, 2016 IEEE Symposium on Service-Oriented System Engineering (SOSE).

[11]  Vijay Mann,et al.  Remedy: Network-Aware Steady State VM Management for Data Centers , 2012, Networking.

[12]  Olivier L. de Weck,et al.  Time-expanded decision networks: A framework for designing evolvable complex systems , 2007 .

[13]  Xavier Hesselbach,et al.  Virtual Network Embedding: A Survey , 2013, IEEE Communications Surveys & Tutorials.

[14]  Wolf-Dietrich Weber,et al.  Power provisioning for a warehouse-sized computer , 2007, ISCA '07.

[15]  Anja Feldmann,et al.  Optimizing Long-Lived CloudNets with Migrations , 2012, 2012 IEEE Fifth International Conference on Utility and Cloud Computing.

[16]  Michal Pioro,et al.  SNDlib 1.0—Survivable Network Design Library , 2010 .

[17]  Christos Kozyrakis,et al.  Full-System Power Analysis and Modeling for Server Environments , 2006 .

[18]  Go Hasegawa,et al.  Reducing Power Consumption in Data Center by Predicting Temperature Distribution and Air Conditioner Efficiency with Machine Learning , 2016, 2016 IEEE International Conference on Cloud Engineering (IC2E).

[19]  Anja Feldmann,et al.  Live wide-area migration of virtual machines including local persistent state , 2007, VEE '07.