Optical data center networks (DCNs) are becoming increasingly attractive due to their technological strengths compared with the traditional electrical networks. However, existing optical DCNs are either hard to scale, vulnerable to single point of failure, or provide limited network bisection bandwidth for many practical data center workloads. To this end, we present WaveCube, a scalable, fault-tolerant, high-performance optical DCN architecture. To scale, WaveCube removes MEMS,1a potential bottleneck, from its design. WaveCube is fault-tolerant, since it does not have single point of failure and there are multiple node-disjoint parallel paths between any pair of top-of-rack switches. WaveCube delivers high performance by exploiting multi-pathing and dynamic link bandwidth along the path. For example, our evaluation results show that, in terms of network bisection bandwidth, WaveCube outperforms prior optical DCNs by up to 400% and is 70%-85% of the ideal non-blocking network (ı.e., theoretical upper bound) under both realistic and synthetic traffic patterns. WaveCube's performance degrades gracefully under failures-it drops 20% even with 20% links cut. WaveCube also holds promise in practice-its wiring complexity is orders of magnitude lower than Fattree, BCube, and c-Through at scale, and its power consumption is 35% of them.

Optical data center networks (DCNs) are becoming increasingly attractive due to their technological strengths compared to traditional electrical networks. However, prior optical DCNs are either hard to scale, vulnerable to single point of failure, or provide limited network bisection bandwidth for many practical DCN workloads. To this end, we present WaveCube, a scalable, fault-tolerant, high-performance optical DCN architecture. To scale, WaveCube removes MEMS1, a potential bottleneck, from its design. Wave-Cube is fault-tolerant since it does not have single point of failure and there are multiple node-disjoint parallel paths between any pair of Top-of-Rack (ToR) switches. WaveCube delivers high performance by exploiting multi-pathing and dynamic link bandwidth along the path. Our extensive evaluation results show that WaveCube outperforms previous optical DCNs by up to 400% and delivers network bisection bandwidth that is 70%–85% of an ideal non-blocking network under both realistic and synthetic traffic patterns. WaveCube's performance degrades gracefully under failures — it drops 20% even with 20% links cut. WaveCube also holds promise in practice — its wiring complexity is orders of magnitude lower than Fattree, BCube and c-Through at large scale, and its power consumption is 35% of them.

Our measurements on production data center traffic together with recently reported results suggest that data center networks suffer from long-lived congestion caused by core network over subscription and unbalanced workload placement. In contrast to traditional traffic engineering approaches that optimize flow routing, in this paper, we explore the opportunity to address the continuous congestion via optimizing VM placement in virtualized data centers. To this end, we present Virtual Knotter, an efficient online VM placement algorithm to reduce congestion with controllable VM migration traffic as well as low time complexity. Our evaluation with both real and synthetic traffic patterns shows that Virtual Knotter performs close to the baseline algorithm in terms of link unitization, with only 5%-10% migration traffic of the baseline algorithm. Furthermore, Virtual Knotter decreases link congestion time by 53% for the production data center traffic.

Wireless frequency resources are often the limiting factor for WLAN throughput. Hence, wireless channel management is needed to mitigate co-channel interference and improve channel reuse efficiency in WLANs, particularly in those with high-density access point deployment. The key challenge is to accurately predict interference and its effect on network performance. This paper presents a new interference prediction model that incorporates realistic signal and traffic measurements, and accurately estimates network performance under alternative channel assignments. Using this model, we develop a channel management algorithm that can adapt to dynamic network situations. Experiments on an indoor testbed of 35 wireless nodes demonstrate that our algorithm significantly outperforms existing channel management techniques, with an average throughput increase of about 30%.

Automatic channel assignment is a key technique to mitigate signal interference and to increase network capacity in multi-channel multi-radio wireless mesh networks. Existing algorithms either can not make efficient use of available radio frequency resources, or have channel assignment coupled with routing and thus lack the practicality in the real world. In this paper, we describe the design of a traffic-aware routing-independent channel assignment algorithm, which aims to make a better trade-off between mesh network capacity improvement and the practicality for channel assignment. Our algorithm does not need any explicit knowledge from the routing protocols. It assumes long-term stable traffic demand and measures the traffic load on each network link. It then assigns channels to the radios based on the measured traffic demand on the network. Extensive ns2 simulations demonstrate that, compared to [8], decoupling channel assignment from routing in our algorithm does not result in significant performance degradation.

IEEE 802.11 based wireless mesh networks have recently attracted much attention as they provide a quick and inexpensive way to extend access to the existing wired networks. A significant problem of such networks is their limited capacity due to signal interference. A frequently proposed approach to increasing mesh network capacity is to equip each mesh node with multiple radios, which can work on multiple non-interfering channels simultaneously. In this paper, we characterize the relationship between the mesh network capacity and the number of non-interfering channels as well as the number of radios per node, assuming that each radio works on a statically assigned channel. Our main result is that, given k non-interfering channels and m radios on each node, the network capacity improvement is bound by O(min (k, m2/t)), where t is the average node degree per neighbor (i.e., the average number of common channels shared by a pair of neighboring nodes). The parameter t and the number of neighbors around each node determine the mesh network’s average node degree, which affects the network connectivity. To validate our analysis, we use simulation results of two channel assignment algorithms and find that both can achieve a constant factor of the theoretical upper bound. Our work offers practical insights for designing and evaluating channel assignment algorithms and for deploying 802.11 wireless mesh networks.

Widely used wireless LAN infrastructure often deploys multiple access points to cover a large area. A user terminal needs to handoff between these access points as the user moves. Unfortunately, the IEEE 802.11 standard introduces large latency during the handoff process (> 300ms), which greatly affects the performance of mobile applications. Worse yet, it only attempts a handoff when wireless signal quality degrades to a point where current connectivity is threatened. From the application point of view, a desirable WiFi handoff scheme should have sufficiently low handoff latency to support even the most demanding realtime applications, make timely handoff decisions to always connect the terminal to the access point of the best signal quality and/or the lowest load, not incur significant extra power consumption at the terminal, and facilitate easy deployment and be backwards compatible with the 802.11 standard. Very few existing proposals have all these properties. This paper presents the design, implementation, and experimental evaluation of a scheme called LeapFrog that, with efficient, proactive WiFi channel probing, achieves all the four properties.

Providing mobility support in the Internet has become increasingly important since more and more users are on the go. To address this need, many schemes, including IETF mobile IP, HAWAII, cellular IP and others, have been developed. However, each of these proposals has its own weakness, such as large handover latency, or relatively high deployment cost. Recently, wireless mesh networks are emerging as a quick-and-cheap solution for wide wireless coverage, a necessary first-step to enable pervasive mobility service. In this paper, we describe the design of a novel network-based local mobility management scheme for wireless mesh networks. Our scheme achieves low-latency handover (usually less than 20 milliseconds of network-layer handover time) when users move from one access router to the next within a network domain. Real-time voice and video applications can thus be supported without any service interruption during handover. Our scheme requires no software upgrade on mobile hosts and can work with future MAC-layer technologies and routing protocols. This further reduces the deployment cost of wireless mesh networks.

Achieving efficient and fair bandwidth allocation while minimizing packet loss in high bandwidth-delay product networks has long been a daunting challenge. Existing end-to-end congestion control (eg TCP) and traditional congestion notification schemes (eg TCP+AQM/ECN) have significant limitations in achieving this goal. While the recently proposed XCP protocol addresses this challenge, XCP requires multiple bits to encode the congestion-related information exchanged between routers and end-hosts. Unfortunately, there is no space in the IP header for these bits, and solving this problem involves a non-trivial and time-consuming standardization process.In this paper, we design and implement a simple, low-complexity protocol, called Variable-structure congestion Control Protocol (VCP), that leverages only the existing two ECN bits for network congestion feedback, and yet achieves comparable performance to XCP, ie high utilization, low persistent queue length, negligible packet loss rate, and reasonable fairness. On the downside, VCP converges significantly slower to a fair allocation than XCP. We evaluate the performance of VCP using extensive ns2 simulations over a wide range of network scenarios. To gain insight into the behavior of VCP, we analyze a simple fluid model, and prove a global stability result for the case of a single bottleneck link shared by flows with identical round-trip times.

This paper generalizes the TCP Vegas congestion avoidance mechanism and uses accumulation , buffered packets of a flow inside network routers, as a congestion measure based on which a family of congestion control schemes can be derived. We call this model Accumulation-based Congestion Control (ACC), which fits into the nonlinear optimization framework proposed by Kelly. The ACC model serves as a reference for packet-switching network implementations. We show that TCP Vegas is one possible scheme under this model. It is well known that Vegas suffers from round trip propagation delay estimation error and reverse path congestion. We therefore design a new Monaco scheme that solves these problems by employing an out-of-band, receiver-based accumulation estimator, with the support of two FIFO priority queues from the (congested) routers. Comparisons between these two schemes demonstrate that Monaco does not suffer from the problems mentioned above and achieves better performance than Vegas. We use ns-2 simulations and Linux implementation experiments to show that the static and dynamic performance of Monaco matches the theoretic results. One key issue regarding the ACC model in general, i.e., the scalability of bottleneck buffer requirement, and a solution using a virtual queueing algorithm are discussed and evaluated.

This paper generalizes the TCP Vegas congestion avoidance mechanism and proposes a model to use accumulation, buffered packets of a flow inside network routers, as a congestion measure on which a family of congestion control schemes can be derived. We call this model accumulation-based congestion control (ACC). We use a bit-by-bit fluid model to define the accumulation concept and develop a general control algorithm, which includes a set of control policies. The new prove its proportional fairness and global stability. The ACC model serves as a reference for packet network implementations. We show that TCP Vegas is one possible scheme, which fits into the ACC model. It is well known that Vegas suffer from round trip propagation delay estimation error and reverse path queuing delay. We therefore design a new scheme called Monaco, which solves these problems by employing an out-of-band receiver-based accumulation estimator, with minimal support from network routers. Analysis and simulations comparisons between Vegas and Monaco demonstrate the effectiveness of the Monaco accumulation estimator. We use ns-2 simulations to show that the static and dynamic performance of Monaco matches the theoretic results. One key issue regarding the ACC model in general, i.e., the scalability of router buffer requirement, is discussed.