Flurries

Abstract

The combination of Network Function Virtualization (NFV) and Software Defined Networking (SDN) allows flows to be flexibly steered through efficient processing pipelines. As deployment of NFV becomes more prevalent, the need to provide fine-grained customization of service chains and flowlevel performance guarantees will increase, even as the diversity of Network Functions (NFs) rises. Existing NFV approaches typically route wide classes of traffic through preconfigured service chains. While this aggregation improves efficiency, it prevents flexibly steering and managing performance of flows at a fine granularity. To provide both efficiency and flexibility, we present Flurries, an NFV platform designed to support large numbers of short-lived lightweight NFs, potentially running a unique NF for each flow. Flurries maintains a pool of Docker container NFs-several thousand on each host-and resets NF memory state between flows for fast reuse. Flurries uses a hybrid of polling and interrupts to improve throughput and latency while allowing multiple NFs to efficiently share CPU cores. By assigning each NF an individual flow or a small set of flows, it becomes possible to dynamically manage the QoS and service chain functionality for flows at a very fine granularity. Our Flurries prototype demonstrates the potential for this approach to run as many as 80,000 Flurry NFs during a one second interval, while forwarding over 30Gbps of traffic, dramatically increasing data plane customizability.

My VM is Lighter (and Safer) than your Container

“…Beyond these services, lightweight virtualization is crucial to a wide range of use cases, including just-in-time instantiation of services [23,26] (e.g., lters against DDoS attacks, TCP acceleration proxies, content caches, etc.) and NFV [41,51], all while providing signicant cost reduction through consolidation and power minimization [46].…”

MMLite

“…NFs can thus directly access packets with no additional copies (i.e, zero copy I/O). To overcome the 100% CPU utilization with DPDK poll mode driver (PMD), we use a hybrid polling-interrupt driven technique to achieve both performance and resource efficiency [73]. We next describe the functional components that developed in C for our prototype.…”

“…Functional Customization: There is a significant body of work on decoupling control and data planes [39,41,42]. Recent works have demonstrated the resiliency of such decoupled services implemented as microservices [22,73]. CoMB [62] demonstrated the composition of NF services into a consolidated middlebox.…”

Rss++

“…Pipelining involves intercore communication; therefore, it is more expensive than sharding that runs-to-completion. Some DPDK-based packet processing frameworks (such as SoftNIC [14], E2 [47], Flurries [61], and NFP [57]) advocate pipelining for some use-cases. However, as network interfaces are currently reaching 400 Gbps and cores' operating frequency have peaked, it becomes clear that pipelining will dedicate multiple cores just for the first part of the pipeline (i.e., packet reception and, in general, a part of the network stack processing).…”