Communication latency has become one of the determining factors for the performance of parallel clusters. To design low-latency network topologies for high-performance computing clusters, we optimize the diameters, mean path lengths, and bisection widths of circulant topologies. We obtain a series of optimal circulant topologies of size $2^5$ through $2^{10}$ and compare them with torus and hypercube of the same sizes and degrees. We further benchmark on a broad variety of applications including effective bandwidth, FFTE, Graph 500 and NAS parallel benchmarks to compare the optimal circulant topologies and Cartesian products of optimal circulant topologies and fully connected topologies with corresponding torus and hypercube. Simulation results demonstrate superior potentials of the optimal circulant topologies for communication-intensive applications. We also find the strengths of the Cartesian products in exploiting global communication with data traffic patterns of specific applications or internal algorithms.
We propose that clusters interconnected with network topologies having minimal mean path length will increase their overall performance for a variety of applications. We approach our heuristic by constructing clusters of up to 36 nodes having Dragonfly, torus, ring, Chvatal, Wagner, Bidiakis and several other topologies with minimal mean path lengths and by simulating the performance of 256-node clusters with the same network topologies. The optimal (or sub-optimal) low-latency network topologies are found by minimizing the mean path length of regular graphs. The selected topologies are benchm…
Datacenters need networks that support both low-latency and high-bandwidth packet delivery to meet the stringent requirements of modern applications. We present Opera, a dynamic network that delivers latency-sensitive traffic quickly by relying on multi-hop forwarding in the same way as expander-graph-based approaches, but provides near-optimal bandwidth for bulk flows through direct forwarding over time-varying source-to-destination circuits. The key to Opera's design is the rapid and deterministic reconfiguration of the network, piece-by-piece, such that at any moment in time the network imp…
We introduce FatPaths: a simple, generic, and robust routing architecture that enables state-of-the-art low-diameter topologies such as Slim Fly to achieve unprecedented performance. FatPaths targets Ethernet stacks in both HPC supercomputers as well as cloud data centers and clusters. FatPaths exposes and exploits the rich ("fat") diversity of both minimal and non-minimal paths for high-performance multi-pathing. Moreover, FatPaths uses a redesigned "purified" transport layer that removes virtually all TCP performance issues (e.g., the slow start), and incorporates flowlet switching, a techni…
Modern network applications demand low-latency traffic engineering in the presence of network failure while preserving the quality of service constraints like delay and capacity. Fast Re-Route (FRR) mechanisms are widely used for traffic re-routing purposes in failure scenarios. Control plane FRR typically computes the backup forwarding rules to detour the traffic in the data plane when the failure occurs. This mechanism could be computed in the data plane with the emergence of programmable data planes. In this paper, we propose a system (called TEL) that contains two FRR mechanisms, namely, T…
In this paper, we consider how to provide fast estimates of flow-level tail latency performance for very large scale data center networks. Network tail latency is often a crucial metric for cloud application performance that can be affected by a wide variety of factors, including network load, inter-rack traffic skew, traffic burstiness, flow size distributions, oversubscription, and topology asymmetry. Network simulators such as ns-3 and OMNeT++ can provide accurate answers, but are very hard to parallelize, taking hours or days to answer what if questions for a single configuration at even m…
We introduce a high-performance cost-effective network topology called Slim Fly that approaches the theoretically optimal network diameter. Slim Fly is based on graphs that approximate the solution to the degree-diameter problem. We analyze Slim Fly and compare it to both traditional and state-of-the-art networks. Our analysis shows that Slim Fly has significant advantages over other topologies in latency, bandwidth, resiliency, cost, and power consumption. Finally, we propose deadlock-free routing schemes and physical layouts for large computing centers as well as a detailed cost and power mo…
5G and Beyond networks promise low-latency support for applications that need to deliver mission-critical data with strict deadlines. However, innovations on the physical and medium access layers are not sufficient. Additional considerations are needed to support applications under different network topologies, and while network setting and data paths change. Such support could be developed at the transport layer, ensuring end-to-end latency in a dynamic network and connectivity environment. In this paper, we present a partial reliability framework, which governs per-packet reliability through…
Increasingly stringent throughput and latency requirements in datacenter networks demand fast and accurate congestion control. We observe that the reaction time and accuracy of existing datacenter congestion control schemes are inherently limited. They either rely only on explicit feedback about the network state (e.g., queue lengths in DCTCP) or only on variations of state (e.g., RTT gradient in TIMELY). To overcome these limitations, we propose a novel congestion control algorithm, PowerTCP, which achieves much more fine-grained congestion control by adapting to the bandwidth-window product …
R&D Lab, Sun* Inc.（Sun*公司研发实验室）；Le Quy Don Technical University（黎贵惇理工大学）
Along with the rapid development in the field of artificial intelligence, especially deep learning, deep neural network applications are becoming more and more popular in reality. To be able to withstand the heavy load from mainstream users, deployment techniques are essential in bringing neural network models from research to production. Among the two popular computing topologies for deploying neural network models in production are cloud-computing and edge-computing. Recent advances in communication technologies, along with the great increase in the number of mobile devices, has made edge-co…
Enabling Reconfiguration-Communication Overlap for Collective Communication in Optical Networks
School of Computer Science and Technology, University of Science and Technology of China (USTC)（科学技术大学计算机科学与技术学院）；Shanghai Innovation Institute（上海创新研究院）；Suzhou Institute for Advanced Research, USTC（先进研究院，USTC）
The fifth-generation mobile evolution enables several transformations on Next Generation Radio Access Networks (NG-RAN). The RAN protocol stack is splitting into eight possible disaggregated options combined into three network units, i.e., Central, Distributed, and Radio. Besides that, further advances allow the RAN software to be virtualized on top of general-purpose vendor-neutral hardware, dealing with the concept of virtualized RAN (vRAN). The disaggregated network units initiatives reach full interoperability based on the Open RAN (O-RAN). The combination of NG-RAN and vRAN results in vNG…
Emerging chips with hundreds and thousands of cores require networks with unprecedented energy/area efficiency and scalability. To address this, we propose Slim NoC (SN): a new on-chip network design that delivers significant improvements in efficiency and scalability compared to the state-of-the-art. The key idea is to use two concepts from graph and number theory, degree-diameter graphs combined with non-prime finite fields, to enable the smallest number of ports for a given core count. SN is inspired by state-of-the-art off-chip topologies; it identifies and distills their advantages for No…
Resource-disaggregated data centre architectures promise a means of pooling resources remotely within data centres, allowing for both more flexibility and resource efficiency underlying the increasingly important infrastructure-as-a-service business. This can be accomplished by means of using an optically circuit switched backbone in the data centre network (DCN); providing the required bandwidth and latency guarantees to ensure reliable performance when applications are run across non-local resource pools. However, resource allocation in this scenario requires both server-level \emph{and} net…
The recent line of research into topology design focuses on lowering network diameter. Many low-diameter topologies such as Slim Fly or Jellyfish that substantially reduce cost, power consumption, and latency have been proposed. A key challenge in realizing the benefits of these topologies is routing. On one hand, these networks provide shorter path lengths than established topologies such as Clos or torus, leading to performance improvements. On the other hand, the number of shortest paths between each pair of endpoints is much smaller than in Clos, but there is a large number of non-minimal …
Hybrid Orchestration of Edge AI and Microservices via Graph-based Self-Imitation Learning
School of Computer Science, South-Central Minzu University（中央民族大学南中央学院）；Key Laboratory of Cyber-Physical Fusion Intelligent Computing, State Ethnic Affairs Commission（国家民族事务委员会网络物理融合智能计算重点实验室）；Hubei Key Laboratory of Smart Internet Technology, School of Electronic Information and Communication, Huazhong University of Science and Technology（湖北智能互联网技术重点实验室，华中科技大学电子信息与通信学院）；Centre for Frontier AI Research, Agency for Science, Technology and Research (A*STAR)（前沿人工智能研究中心，科技研究局（A*STAR））
Reinforcement Learning for Opportunistic Routing in Software-Defined LEO-Terrestrial Systems
School of Electrical and Mechanical Engineering, University of Adelaide（阿德莱德大学电气与机械工程学院）；Singapore University of Technology and Design（新加坡科技设计大学）；Electronics and Telecommunications Research Institute（电子电信研究院）
