Sharing Cloud Networks with FairCloud

The network is a crucial resource in cloud computing, but in contrast to other resources such as CPU or memory, the network is currently shared in a best effort manner. However, sharing the network in a datacenter is more challenging than sharing other resources. The key difficulty is that the network allocation for a VM X depends not only on the VMs running on the same machine with X, but also on the other VMs that X communicates with, as well as on the crosstraffic on each link used by X. In this paper, we first propose a set of desirable properties for allocating cloud networks with fairness and show that there exist multiple tradeoffs between some of these properties. Second, we show that the existing allocation models violate one or more of these properties, and propose a flexible mechanism that can be tuned to select different points in this tradeoff space.