
doi: 10.1109/ancs.2017.14
In its current form, OpenFlow, the de facto implementation\ud of SDN, separates the network’s control and data\ud planes allowing a central controller to alter the matchaction\ud pipeline using a limited set of fields and actions.\ud To support new protocols, forwarding logic, telemetry,\ud monitoring or even middlebox-like functions the currently\ud available programmability in SDN is insufficient.\ud In this paper, we introduce BPFabric, a platform, protocol,\ud and language-independent architecture to centrally\ud program and monitor the data plane. BPFabric leverages\ud eBPF, a platform and protocol independent instruction\ud set to define the packet processing and forwarding functionality\ud of the data plane. We introduce a control plane\ud API that allows data plane functions to be deployed onthe-fly,\ud reporting events of interest and exposing network\ud internal state.\ud We present a raw socket and DPDK implementation\ud of the design, the former for large-scale experimentation\ud using environment such as Mininet and the latter for\ud high-performance low-latency deployments. We show\ud through examples that functions unrealisable in OpenFlow\ud can leverage this flexibility while achieving similar\ud or better performance to today’s static design.
| selected citations These citations are derived from selected sources. This is an alternative to the "Influence" indicator, which also reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically). | 21 | |
| popularity This indicator reflects the "current" impact/attention (the "hype") of an article in the research community at large, based on the underlying citation network. | Top 10% | |
| influence This indicator reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically). | Top 10% | |
| impulse This indicator reflects the initial momentum of an article directly after its publication, based on the underlying citation network. | Top 10% |
