27 February 2023 in Proceedings of the ACM on Measurement and Analysis of Computing Systems
Programmable switches have been widely used to design network monitoring solutions that operate in the fast data-plane level, e.g., detecting heavy hitters, super-spreaders, computing flow size distributions and their entropy. Many existing works on networking monitoring assume switches deploy a single memory that is accessible by each processed packet. However, high-speed ASIC switches increasingly deploymultiple independent pipes, each equipped with its own independent memory thatcannot be accessed by other pipes. In this work, we initiate the study of deploying existing heavy-hitter data-plane monitoring solutions on multi-pipe switches where packets of a "flow" may spread over multiple pipes, i.e., stored into distinct memories. We first quantify the accuracy degradation due to splitting a monitoring data structure across multiple pipes (e.g., up to 3000x worse flow-size estimation average error). We then present PipeCache, a system that adaptsexisting data-plane mechanisms to multi-pipe switches by carefully storing all the monitoring information of each traffic class into exactly one specific pipe (as opposed to replicate the information on multiple pipes). PipeCache relies on the idea of briefly storing monitoring information into a per-pipe cache and then piggybacking this information onto existing data packets to the correct pipeentirely at data-plane speed. We implement PipeCache on ASIC switches and we evaluate it using a real-world trace. We show that existing data-plane mechanisms achieves accuracy levels and memory requirements similar to single-pipe deployments when augmented with PipeCache (i.e., up to 16x lower memory requirements).
19 January 2022 in Journal of Systems and Software
10 April 2020 in IEEE Transactions on Network and Service Management
HOHelio Tibagi de Oliveira
Over the years, network monitoring tools have evolved to keep pace with the advances in networking technologies and paradigms, such as virtualization and softwarization, and more recently, data plane programmability. However, there have been few advances regarding user interfaces and interactions in network management systems. In this paper, an investigation based on the human-computer interaction perspective is presented and improvements on the usability aspects of network monitoring tools are proposed. First, we conducted a survey to capture general information from network administrator professionals. On the basis of these initial data, we identified and specified two personas to represent the groups of users. Thereafter, a study focused on observing users’ interaction and experience with a well-known network monitoring tool was conducted. Finally, our qualitative analysis revealed several findings that were used to develop recommendations to aid developers in the (re)design of these tools to enhance end-user interaction.
6 June 2023 in IEEE Transactions on Network and Service Management
InFaRR (In-network Fast ReRouting) is an algorithm for fast rerouting in programmable data planes. Implemented in P4, InFaRR is free of additional headers and network state heartbeats. InFaRR has four essential features not jointly found in other recovery mechanisms: Loop prevention, Pushback, Recognition and Restoration and Return to the main route. Tests in standard Fat-Tree and AB Fat-Tree topologies with failures in different scenarios showed positive results when compared to state-of-the-art algorithms in the literature. In scenarios in which the other algorithms were able to recover, InFaRR presented fewer hops to bypass the failure when the Pushback, Loop Prevention and Recognition and Restoration mechanisms were used. In scenarios with multiple failures, InFaRR successfully rerouted where the others algorithms in some cases looped. The mechanism for returning to the main route allows to verify failures in remote links, making it possible to return to the main route without intervention from the control plane. Several evaluations were done comparing the results of InFaRR with the state-of-the art mechanisms, showing the capability of our fast rerouting algorithm in dealing with failures at the line-rate.