Future SLA-Aware Clouds: Bandwidth-aware management of virtual machines (Future X Days 2014 Stuttgart)

Bell Labs has developed new mechanisms to consider network constraints of Virtual Machines (VMs) in cloud management systems. These mechanisms enable a network-aware placement of instances that takes into account bandwidth constraints to and from compute nodes in addition to computing resources. The mechanisms have been implemented based on the open source cloud computing system OpenStack, using the "Icehouse" release. The demo shows OpenStack extensions for network-aware placement intelligence in the OpenStack Nova scheduler. In the proof-of-concept we demonstrate the benefit of the Bell Labs innovation compared to the OpenStack default solution. Our realization is flexible and can handle different types of instances and bandwidth requirements. It tracks the utilization of resources resource and leverages network monitoring data, and it can also be combined with mechanisms for SLA enforcement such as rate limiting.