E2

By moving network appliance functionality from proprietary
\nhardware to software, Network Function Virtualization
\npromises to bring the advantages of cloud computing to
\nnetwork packet processing. However, the evolution of cloud
\ncomputing (particularly for data analytics) has greatly bene-
\nfited from application-independent methods for scaling and
\nplacement that achieve high efficiency while relieving programmers
\nof these burdens. NFV has no such general management
\nsolutions. In this paper, we present a scalable and
\napplication-agnostic scheduling framework for packet processing,
\nand compare its performance to current approaches.