Wenxue Cheng

In cloud datacenter operations, telemetry and logs are indispensable, enabling essential services such as network diagnostics, auditing, and knowledge discovery. The escalating scale of data centers, coupled with increased bandwidth and finer-grained telemetry, results in an overwhelming volume of data. This proliferation poses significant storage challenges for telemetry systems. In this paper, we introduce HyperDrive, an innovative system designed to efficiently store large volumes of telemetry and logs in data centers using programmable switches. This in-network approach effectively mitigates bandwidth bottlenecks commonly associated with traditional endpoint-based methods. To our knowledge, we are the first to use a programmable switch to directly control storage, bypassing the CPU to achieve the best performance. With merely 21% of a switch's resources, our HyperDrive implementation showcases remarkable scalability and efficiency. Through rigorous evaluation, it has demonstrated linear scaling capabilities, efficiently managing 12 SSDs on a single server with minimal host overhead. In an eight-server testbed, HyperDrive achieved an impressive throughput of approximately 730Gbps, underscoring its potential to transform data center telemetry and logging practices.

Unlike traditional data center traffic, AI training traffic primarily consists of large-size flows that are fewer in number. This characteristic poses a challenge in balancing routing granularity with reorder overhead in existing routing strategies. Existing serial flowlet schemes aim to achieve a better trade-off in TCP scenarios than flow-level or packet spraying load balancing. However, they are not well-suited for AI training clusters with high-performance RDMA networks. To tackle this issue, we propose a parallel-flowlet strategy, ParaLet, which effectively resolves the serial flowlet's problems of insufficient routing entropy in AI training traffic and the difficulty of identifying time gaps in RDMA networks. ParaLet requires only a small number of Queue Pairs, which are decoupled from the connections, thus circumventing scalability limits. The theoretical analysis and simulations indicate that ParaLet not only achieves near-optimal throughput but also diminishes flow completion time by 1.5-3.4 times compared to existing methods.

With the growing importance of real-time communications (RTC), designing congestion control (CC) algorithms for RTC that achieve high network performance and QoE is gaining attention. Recently, data-driven, reinforcement learning (RL)-based CC algorithms for RTC have shown great potential, outperforming traditional rule-based counterparts. However, there are no open platforms tailored for training, evaluation, and validation of the algorithms that can facilitate this emerging research area. We present OpenNetLab, an open platform for fast training, reproducible end-to-end evaluation, and performance validation of RL-based CC algorithms for RTC. Preliminary use cases confirm that OpenNetLab concretely aided the training of novel RL-based CC algorithms for RTC that outperform a well-established rule-based baseline in both network performance and QoE metrics.

Recently, the 40-100Gbps Energy Efficient Ethernet (EEE) is standardized by IEEE 802.3bj. Unlike the 1-10Gbps EEE, the 40-100Gbps EEE includes two energy-saving states with different power consumption and state transition times. The default Dual-Mode strategy is the fundamental policy, which utilizes these states to make a trade-off between the energy saving and the incurred latency of frames. Although numerous analytical works have been proposed for EEE strategies, most of them can only be applied to the single energy-saving state condition of 1-10Gbps EEE, or can only capture the mean latency of frames. However, the interactive applications always place a soft or hard deadline on frame transmission. Accordingly, the tail latency of frames becomes crucial. Therefore, we model and analyze the latency distribution of frames in the 40-100Gbps Dual-Mode EEE in this work. The guidelines provided by analytical results help the Dual-Mode strategy to fulfill the given tail latency requirement and make a better trade-off between the energy saving and the incurred latency under different traffic loads. In addition, this model is powerful as it can generate the major results of several existing works under the corresponding simplified scenarios. Finally, the analytical results are verified by trace-driven simulations based on NS3.

Memcached is a widely used in-memory caching solution in large-scale searching scenarios. The most crucial metric of Memcached systems is the response time, which is affected by various factors such as workload, service rate, unbalanced load distribution, and cache miss ratio. This paper aims to quantify the influence of each factor on the response time of Memcached systems. First, we establish a theoretical model for Memcached systems that captures their main features, including burst and concurrent key arrival, unbalanced load distribution and cache miss process. By solving this model using queuing and stochastic theories, we obtain an estimate of the response time in Memcached systems. Intensive experiments based on real-world components demonstrate that the estimate always matches perfectly with the actual value. Furthermore, we obtain a comprehensive and quantitative understanding of all factors. The main insights are threefold. 1) There exists an optimum range of utilization at Memcached servers in which the response time is kept at a low level with a small penalty. 2) The influence of the cache miss ratio on the response time is logarithmic rather than linear. 3) The number of keys generated from an end-user request has the greatest impact in Memcached systems.

In recent years, the growth in popularity of mobile video streaming services is unbroken. There are tremendous demands for video streaming over wireless networks. Currently, most video streaming is over HTTP. Up to now, HTTP-based adaptive video streaming is standardized as DASH, where a client-side video player can dynamically pick the bitrate level according to the perceived network conditions. Actually, not only the available bandwidth drastically varies due to wireless network properties, but also the chunk sizes in the same bitrate level significantly fluctuate, which also influences the bitrate adaptation. However, existing bitrate adaptation algorithms mostly focus on available bandwidth but do not involve chunk size variation, leading to performance losses. In this paper, we theoretically analyze the influence of chunk size variation on bitrate adaptation performance in wireless networks. Based on DASH system features, we build a general model describing playback buffer evolution. Applying stochastic theories, we respectively analyze the influence of the chunk size variation on rebuffering probability, average bitrate, and bitrate switching interval. Furthermore, based on theoretical insights, we provide several suggestions for algorithm designing and rate encoding, and also design a simple bitrate adaptation algorithm. Extensive simulations verify our insights, suggestions and designed algorithm effectiveness.

Overload detecting mechanisms are the prerequisites for admission controls and differentiated services. Conventional mechanisms are always coordinated with conventional network stacks. As the hardware evolution and applications' increasing demand, the architecture of conventional network stacks make themselves become the bottleneck. In this context, high performance network I/O frameworks have been proposed and adopted by both academia and industry. However, the characteristics of these frameworks limit the effectiveness of conventional overload detecting mechanisms. In this paper, we propose that RX RING length can act as a good threshold signal to reflect whether overload happens or not. Through experiments with varying the input load and the processing task load, RX RING length is proved to be a high-quality indicator for overload detecting. Meanwhile, we design a low-cost algorithm to detect RX RING length in real time. Compared with the original method, our algorithm reduces more than 90% of running cost when RX RING is nearly full. This makes RX RING possible to be a high-quality indicator in future overload detecting mechanism design for applications based high performance network I/O frameworks.