Cross-silo Federated Learning (FL) enables multiple institutions to collaboratively train machine learning models while preserving data privacy. In such settings, clients repeatedly exchange model weights with a central server, making the overall training time highly sensitive to network performance. However, conventional routing methods often fail to prevent congestion, leading to increased communication latency and prolonged training. Software-Defined Networking (SDN), which provides centralized and programmable control over network resources, offers a promising way to address this limitation. To this end, we propose SmartFLow, an SDN-based framework designed to enhance communication efficiency in cross-silo FL. SmartFLow dynamically adjusts routing paths in response to changing network conditions, thereby reducing congestion and improving synchronization efficiency. Experimental results show that SmartFLow decreases parameter synchronization time by up to 47% compared to shortest-path routing and 41% compared to capacity-aware routing. Furthermore, it achieves these gains with minimal computational overhead and scales effectively to networks of up to 50 clients, demonstrating its practicality for real-world FL deployments.

Federated Learning (FL) presents a robust paradigm for privacy-preserving, decentralized machine learning. However, a significant gap persists between the theoretical design of FL algorithms and their practical performance, largely because existing evaluation tools often fail to model realistic operational conditions. Many testbeds oversimplify the critical dynamics among algorithmic efficiency, client-level heterogeneity, and continuously evolving network infrastructure. To address this challenge, we introduce the Federated Learning Emulation and Evaluation Testbed (FLEET). This comprehensive platform provides a scalable and configurable environment by integrating a versatile, framework-agnostic learning component with a high-fidelity network emulator. FLEET supports diverse machine learning frameworks, customizable real-world network topologies, and dynamic background traffic generation. The testbed collects holistic metrics that correlate algorithmic outcomes with detailed network statistics. By unifying the entire experiment configuration, FLEET enables researchers to systematically investigate how network constraints, such as limited bandwidth, high latency, and packet loss, affect the convergence and efficiency of FL algorithms. This work provides the research community with a robust tool to bridge the gap between algorithmic theory and real-world network conditions, promoting the holistic and reproducible evaluation of federated learning systems.

Modern storage-intensive server applications such as key-value stores and relational databases often rely on thread oversubscription to sustain high throughput in cloud environments. While effective at hiding I/O stalls, this practice introduces serious challenges, including unpredictable query behaviors, instruction-per-query (IPQ) inflation, and instability in dynamic power management (DPM). Existing quality-of-service (QoS)-centric or energy-centric techniques, developed in isolation at one of the service layers, fail to holistically optimize resource and energy efficiency under connection-level QoS constraints. This paper presents REEF (Resource- and Energy-Efficient user-space scheduling Framework), a non-intrusive, cross-layer approach that coordinates query processing across the network stack, application layer, and OS resource manager. REEF transforms self-serving threads into on-call threads activated by near-optimal, proactively batched scheduling, enabling deep CPU C-state residency and mitigating CPU and I/O contention. Our extensive evaluation on real server applications (MongoDB and MySQL) demonstrates that REEF can substantially improve the energy efficiency of server applications under different connection-level QoS schemes, by up to 53.45% for throughput per power, up to 2.18× and 5.33× for coefficient of P99 and P99.9 tail-latency per power respectively, and significantly reduce resource consumption, by up to 73.65% in CPU frequency.

A fundamental design principle of MultiPath TCP (MPTCP) congestion control algorithm (CCA) is that an MPTCP flow should be fair to and do not harm TCP flows. Unfortunately, to deal with cost heterogeneity among subflow interfaces, the existing cost-aware MPTCP CCAs often violate this design principle in an attempt to minimize the cost. Based on the network utility maximization (NUM) framework, we put forward UniMPTCP($\vec \omega $,n), a NUM-optimal, Unified MPTCP CCA with n subflow paths and a n-dimension weight vector $\vec \omega $ with n −1 independent elements. Uni-MPTCP($\vec \omega $,n) abides by this design principle for arbitrary $\vec \omega $ and can be customized to achieve specific cost design objectives with proper adaptation of $\vec \omega $. As such, UniMPTCP($\vec \omega $,n) provides a unified solution to enable cost-aware MPTCP CCAs, while adhering to the design principle. Finally, we put forward an adaptation algorithm for, ω, in Uni-MPTCP(ω,2), aiming at maintaining a target MPTCP flow rate with minimum cost for a cost-heterogeneity case with dual connectivity. The test results based on NS-3 simulation demonstrate that UniMPTCP(ω,2) can indeed effectively keep track of a given flow rate target with minimum cost, while adhering to the design principle.

A primary design objective for user-facing services for cloud and edge computing is to maximize query throughput, while meeting query tail latency Service Level Objectives (SLOs) for individual queries. Unfortunately, the existing solutions fall short of achieving this design objective, which we argue, is largely attributed to the fact that they fail to take the query fanout explicitly into account. In this paper, we propose TailGuard based on a Tail-latency-SLO-and-Fanout-aware Earliest-Deadline-First Queuing policy (TF-EDFQ) for task queuing at individual task servers the query tasks are fanned out to. With the task pre-dequeuing time deadline for each task being derived based on both query tail latency SLO and query fanout, TailGuard takes an important first step towards achieving the design objective. A query admission control scheme is also developed to provide tail latency SLO guarantee in the presence of resource shortages. TailGuard is evaluated against First-In-First-Out (FIFO) task queuing, task PRIority Queuing (PRIQ) and Tail-latency-SLO-aware EDFQ (T-EDFQ) policies by both simulation and testing in the Amazon EC2 cloud. It is driven by three types of applications in the Tailbench benchmark suite, featuring web search, in-memory key-value store, and transactional database applications. The results demonstrate that TailGuard can significantly improve resource utilization (e.g., up to 80% compared to FIFO), while also meeting the targeted tail latency SLOs, as compared with the other three policies. TailGuard is also implemented and tested in a highly heterogeneous Sensing-$a$s-a-Service (SaS) testbed for a data sensing service, demonstrating performance gains of up to 33% . These results are consistent with both the simulation and Amazon EC2 results.

Federated computing, including federated learning and federated analytics, needs to meet certain task Service Level Objective (SLO) in terms of various performance metrics, e.g., mean task response time and task tail latency. The lack of control and access to client activities requires a carefully crafted client selection process for each round of task processing to meet a designated task SLO. To achieve this, one must be able to predict task performance metrics for a given client selection per round of task execution. In this paper, we develop, FedSLO, a general framework that allows task performance in terms of a wide range of performance metrics of practical interest to be predicted for synchronous federated computing systems, in line with the Google federated learning system architecture. Specifically, with each task performance metric expressed as a cost function of the task response time, a relationship between the task performance measure - the mean cost and task/subtask response time distributions is established, allowing for unified task performance prediction algorithms to be developed. Practical issues concerning the computational complexity, measurement cost and implementation of FedSLO are also addressed. Finally, we propose preliminary ideas on how to apply FedSLO to the client selection process to enable task SLO guarantee.

Unexpected long query latency of a database system can cause domino effects on all the upstream services and severely degrade end users' experience with unpredicted long waits, resulting in an increasing number of users disengaged with the services and thus leading to a high user disengagement ratio (UDR). A high UDR usually translates to reduced revenue for service providers. This paper proposes UTSLO, a UDR-oriented SLO guaranteed system, which enables a database system to support multi-tenant UDR targets in a cost-effective fashion through UDR-oriented capacity planning and dynamic UDR target enforcement. The former aims to estimate the feasibility of UDR targets while the latter dynamically tracks and regulates per-connection query latency distribution needed for accurate UDR target guarantee. In UTSLO, the database service capacity can be fully exploited to efficiently accommodate tenants while minimizing resources required for UDR target guarantee.

Cross-speaker style transfer in speech synthesis aims at transferring a style from source speaker to synthesized speech of a target speaker’s timbre. In most previous methods, the synthesized fine-grained prosody features often represent the source speaker’s average style, similar to the one-to-many problem(i.e., multiple prosody variations correspond to the same text). In response to this problem, a strength-controlled semi-supervised style extractor is proposed to disentangle the style from content and timbre, improving the representation and interpretability of the global style embedding, which can alleviate the one-to-many mapping and data imbalance problems in prosody prediction. A hierarchical prosody predictor is proposed to improve prosody modeling. We find that better style transfer can be achieved by using the source speaker’s prosody features that are easily predicted. Additionally, a speaker-transfer-wise cycle consistency loss is proposed to assist the model in learning unseen style-timbre combinations during the training phase. Experimental results show that the method outperforms the baseline. We provide a website with audio samples1.

A primary design objective for Data-intensive User-facing (DU) services for cloud and edge computing is to maximize query throughput, while meeting query tail latency Service Level Objectives (SLOs) for individual queries. Unfortunately, the existing solutions fall short of achieving this design objective, which we argue, is largely attributed to the fact that they fail to take the query fanout explicitly into account. In this paper, we propose TailGuard based on a Tail-latency-SLO-and-Fanout-aware Earliest-Deadline-First Queuing policy (TF-EDFQ) for task queuing at individual task servers the query tasks are fanned out to. With the task queuing deadline for each task being derived based on both query tail latency SLO and query fanout, TailGuard takes an important first step towards achieving the design objective. TailGuard is evaluated against First-In-First-Out (FIFO) task queuing, task PRIority Queuing (PRIQ) and Tail-latency-SLO-aware EDFQ (T-EDFQ) policies by simulation. It is driven by three types of applications in the Tailbench benchmark suite. The results demonstrate that TailGuard can improve resource utilization by up to 80%, while meeting the targeted tail latency SLOs, as compared with the other three policies. TailGuard is also implemented and tested in a highly heterogeneous Sensing-as-a-Service (SaS) testbed for a data sensing service, with test results in line with the other ones.

Modern database systems suffer from compromised throughput, persistent unfair I/O processing and unpredictable, high latency variability of user requests as a result of mismatches between highly scaled user parallel I/O and the I/O capacity afforded by the database and its underlying storage I/O stack. To address this problem, we introduce an efficient user-centric QoS-aware scheduling shim, called AppleS, for user-level fine-grained I/O regulation that delivers the right amount and pattern of user parallel I/O requests to the database system and supports user SLOs with high-level performance isolation and reduced I/O resource contention. It is designed to enable database systems to proactively regulate user request behaviors based on runtime conditions to reshape user access pattern to hide excessive user parallelism from the I/O stack that has a limited concurrent processing capability. This helps achieve scalable throughput for multi-user workloads in a fair and stable manner. AppleS is implemented as a user-space shim for transparent user-differentiated I/O scheduling, making it highly flexible and portable. Our extensive evaluation, run on real databases (MySQL and MongoDB), demonstrates that, by incorporating AppleS in the existing database systems, our solution can not only improve the throughput (up to 39.2%) in a fairer (3.2× to 40.6× fairness improvement) and more stable (up to 2× lower latency variability) manner, but also support user SLOs with less I/O provisioning.

Singing voice conversion (SVC) converts a singer’s voice to another one’s voice while preserving the linguistic content. Recently, some SVC systems rely on supervised phonetic features extracted from pre-trained automatic speech recognition (ASR) models, increasing system complexity. Some end-toend SVC systems use adversarial training, which causes instability during optimization. To address these issues, we present K-Converter, a simple system to disentangle the timbre, pitch, and content information without any manual supervision or adversarial training. First, low quefrencies of mel-frequency cepstral coefficients (MFCC), which remove the glottal excitation mainly, are used as input representations. And the pitch-shift augmentation is used for further disentangling the pitch. Second, an encoder network is carefully designed to construct an information bottleneck, which learns to break up the pitch and timbre information of the source. Third, the content consistency loss is introduced to keep the content consistent between encoder outputs of source utterances and reconstructed ones. Experimental results show that our proposed system performs well in both speech naturalness and timbre similarity, with better robustness to comparisons.

Multiple Path Transmission Control Protocols (MPTCPs) allow flows to explore path diversity of datacenter networks and multihoming to improve throughput, reliability, and network resource utilization. However, the existing MPTCPs are largely empirical by design and fall short of achieving satisfactory tradeoffs among responsiveness, TCP fairness and throughput. By leveraging the TCP utility and a network-utility-maximization (NUM) solution for concave utilities, in this paper, we derive and implement in Linux kernels two distinct families of NUM-optimal MPTCP protocols, EUTCP ($\gamma$), a function of a rate-scaling vector of the sub-flow rate-scaling coefficients,$\gamma$, and WUTCP ($\omega$), a function of a utility weight vector of the sub-flow weights,$\omega$, respectively. While the former allows resource pooling, the latter does not. We then show that the Semicoupled algorithm and EWTCP are in fact EUTCP(1) and WUTCP$(1/m^{2})$, where$m$is the number of sub-flow paths, and hence, are NUM-optimal. The performance of the two families with equal weight and equal rate-scaling coefficient for all sub-flows when coexisting with TCP is also analyzed based on experiments in a testbed. In particular, the test results demonstrate that the family members of EUTCP ($\gamma$) with rate-scaling coefficient in the range of [1, 1.1] outperform three well-known MPTCPs with resource pooling capability, including LIA, OLIA and Balia, in terms of achieving satisfactory tradeoffs among responsiveness, fairness and throughput. Finally, the effectiveness of the proposed algorithms compared to the existing ones is further confirmed by simulation in a fat-tree datacenter network topology running both long and short flows.

The Fork-Join structure underlays many distributed computing applications in data centers. In this paper, we develop a technique, called ForkMV, to estimate the mean and variance of request response time for Fork-Join queuing networks (FJQNs) with both short-tailed and long-tailed service time distributions and arbitrary request fanout degrees (i.e., the number of Fork nodes). Specifically, for an FJQN with any given service time distribution of practical interests, ForkMV is able to estimate the mean and variance of request response time, accurate enough to facilitate effective resource allocation for data center applications. The test results indicate that in the entire range of the fanout degrees being tested (i.e., [1], [4000]), ForkMV is able to estimate the mean response time within 5 % and 15 % and variance response time within 15% and 10% of the simulation results for short-tailed exponential service distribution and long-tailed truncated Pareto distribution, respectively, at the 90 % load or higher.

Elasticity is an essential feature of cloud computing, which allows users to dynamically add or remove resources in response to workload changes. However, building applications that truly exploit elasticity is non-trivial. Traditional applications need to be modified to efficiently utilize variable resources. This paper explores thread oversubscription, i.e., provisioning more threads than the available cores, to exploit CPU elasticity in the cloud. While maintaining sufficient concurrency allows applications to utilize additional CPUs when more are made available, it is widely believed that thread oversubscription introduces prohibitive overheads due to excessive context switches, loss of locality, and contention on shared resources.

Voice conversion (VC) is a task to convert the voice of speech while preserving its linguistic content. Although several methods have been proposed to enable VC with non-parallel data, it is still difficult to model the voice without a great number of data or an adaptive process. In this paper, we propose a speaker-aware voice conversion (SAVC) system realizing one-shot voice conversion without an adaptation stage. The SAVC utilizes a speaker aware module (SAM) to disentangle speaker embeddings. The SAM comprises a dynamic reference encoder, a static speaker knowledge block (SKB), and a multi-head attention layer. The reference encoder is used to compress a variable-length utterance to a fixed-length vector, the SKB is made up of pre-extraction x-vectors, and the multi-head attention layer is designed to generate weighted combined speaker embeddings. Subsequently, phonetic pos- teriorgrams (PPGs) as context encoding are concatenated with speaker embeddings and sent to the decoder module for generating acoustic features. Experimental results on the Aishell-1 corpus show that the proposed method can improve speaker similarity and converted utterances' speech quality.

TCAM is widely used for flow table lookup in Software-Defined Networking (SDN) switches for datacenter and enterprise networks. While its lookup throughput is unparalleled, TCAM updating, particularly for new rule insertions, can impair the overall system performance. A rule insertion entails two steps: 1) Computing the rule moving operations; and 2) Interrupting the TCAM lookups to apply the operations. In previous work, the performance gain on one step is always at the expense of the performance loss on the other. However, update throughput and latency depend on both. In this paper, we present a faster and more balanced TCAM update scheme, which not only achieves the shortest interrupt time so far but also significantly reduces the computation time. By using a novel sequential stack, FastUp reduces the time and space complexity of the state-of-the-art schemes from$O(m^{2})$and$O(m)$to$O(m\log h)$and$O(h)$, respectively, where$h << m$. Evaluations show that FastUp shortens the computation time and the interrupt time by$100\times$and$1.6\times$, respectively, which is equivalent to update delay${15\times}$reduction and$\mathbf{10\times}$update throughput gain against the state-of-the-art schemes. Moreover, we debunk a common mistake and show the dynamic programming based algorithm cannot be used to solve the reorder problem, and instead we use a bidirectional rule moving method to address the problem. In addition, we propose a practical method to find the theoretical lower bound of interrupt time in relatively large TCAM, which can be used to evaluate the optimality degree of TCAM update schemes. Evaluations show that FastUp achieves 90 % optimality.

Today s datacenters need to meet service level objectives (SLOs) for applications, which can be translated into deadlines for (co)flows running between job execution stages. As a result, meeting (co)flow deadlines with high probabilities is essential to attract and retain customers and hence, generate high revenue. To fill the lack of a transport protocol that can facilitate low (co)flow deadline miss rate, especially in the face of incast congestion, in this paper, we propose DCMRG, an incast-coflow-aware, ECN-based soft minimum-rate-guaranteed congestion control protocol for datacenter applications. DCMRG is composed of two major components, i.e., a congestion controller running on the send host and an incast congestion controller running on the receive host. DCMRG possesses three salient features. First, it is the first congestion control protocol that integrates congestion control with coflow-aware incast control while providing soft minimum flow rate guarantee. Second, DCMRG is readily deployable in datacenter networks. It only requires software upgrade in the hosts and minimum assistance (i.e., ECN) from in-network nodes. Third, DCMRG is backward compatible with and, by design, friendly to the widely deployed, standard-based transport protocols, such as DCTCP. The results from large-scale datacenter network simulation demonstrate that in the absence of incast congestion, DCMRG can reduce flow deadline miss rates by 3x and 1.6x compared to D2TCP and MRG, respectively. Moreover, DCMRG further reduces the coflow deadline miss rate by more than 40% and 60% and lowers the packet drop probability by 60% and 80%, in the face of incast congestion, compared to D2TCP with ICTCP and MRG with ICTCP, respectively.

While widely used for flow tables in SDN switches, TCAM faces challenges for rule updates. Both the computation time and interrupt time need to be short. We propose FastUp, a new TCAM update algorithm, which improves the previous dynamic programming-based algorithms. Evaluations show that FastUp shortens the computation time by 40~100× and the interrupt time by 1.2~2.5×. In addition, we are the first to prove the NP-hardness of the optimal TCAM update problem, and provide a practical method to evaluate an algorithm's degree of optimality. Experiments show that FastUp's optimality reaches 90%.

In this paper, we put forward a HOListic traffic control framework for datacenter NETworks (HOLNET). HOLNET reformulates the network utility maximization (NUM) framework into a HOLNET NUM framework that fully harnesses the potential of the existing NUM-based solutions to allow large families of traffic control protocols of various degrees of sophistication to be developed, i.e., host-based, single or multiple Class-of-Service (CoS) enabled, single or multi-path congestion control, with or without in-network load balancing. Unlike the existing solutions that are largely empirical and point by design, HOLNET is a principled, systematic framework. All the protocols in a family developed under HOLNET share a common, user-defined global optimization objective and fairness criterion. As a result, the protocols in a family can be fairly compared and carefully selected to fully explore the performance, scalability and design complexity tradeoffs. Case studies, based on both a single and a multi-path host-based solutions, demonstrate the viability and flexibility in HOLNET design space exploration. To further test the backward compatibility and performance with respect to some existing lightweight solutions, we develop HOLNET-UTA, an integrated congestion control and load balancing protocol, achieving TCP-fair resource allocation. HOLNET-UTA is found by simulation to improve the average flow completion time (FCT) by more than 20%, compared to DRILL with DCTCP.

RAID-6 is gradually replacing RAID-5 as the dominant form of disk arrays due to its capability of tolerating concurrent failures of any two disks, as well as the case of encountering an uncorrectable read error during recovery. Implementing a RAID-6 system relies on some erasure coding schemes, and so far the most representative solutions are EVENODD codes [1], RDP codes [2] and Liberation codes [3], none of which has emerged as a clear "all-around" winner. In this paper, we are interested in revealing the undiscovered potential of the Liberation codes, since these codes have the following attractive features: (a) they have the best update performance, (b) they have better scalability, and (c) they are open-sourced and publicly available, as well as the following drawbacks: fair encoding performance and, more importantly, relatively poor decoding performance. Specificly, we present novel optimal encoding and decoding algorithms for the Liberation codes by introducing an alternative, geometric presentation of these codes. The proposed algorithms completely eliminate redundant computations during the encoding and decoding procedures by extracting and reusing common expressions between the two types of parity constraints, and do not involve any matrix operations on which the original algorithms are based. Our experiment results show that compared with the original solution, the proposed encoding and decoding algorithms reduce the number of XOR's by up to 16 percent and 15 ~20 percent respectively, and the encoding and decoding throughputs are increased by 22.3 percent and at most 155 percent respectively. Moreover, the encoding complexity reaches the theoretical lower bound, while the decoding complexity is also very close to the theoretical lower bound.

The workflows of the predominant datacenter services are underlaid by various Fork-Join structures. Due to the lack of good understanding of the performance of Fork-Join structures in general, today's datacenters often operate under low resource utilization to meet stringent service level objectives (SLOs), e.g., in terms of tail and/or mean latency, for such services. Hence, to achieve high resource utilization, while meeting stringent SLOs, it is of paramount importance to be able to accurately predict the tail and/or mean latency for a broad range of Fork-Join structures of practical interests. In this article, we propose a black-box Fork-Join model that covers a wide range of Fork-Join structures for the prediction of tail and mean latency, called ForkTail and ForkMean, respectively. We derive highly computational effective, empirical expressions for tail and mean latency as functions of means and variances of task response times. Our extensive testing results based on model-based and trace-driven simulations, as well as a real-world case study in a cloud environment demonstrate that the models can consistently predict the tail and mean latency within 20 and 15 percent prediction errors at 80 and 90 percent load levels, respectively, for heavy-tailed workloads, and at any load levels for light-tailed workloads. Moreover, our sensitivity analysis demonstrates that such errors can be well compensated for with no more than 7 percent resource overprovisioning. Consequently, the proposed prediction model can be used as a powerful tool to aid the design of tail-and-mean-latency guaranteed job scheduling and resource provisioning, especially at high load, for datacenter applications.

In today's datacenters, job heterogeneity makes it difficult for schedulers to simultaneously meet latency requirements and maintain high resource utilization. The state-of-the-art datacenter schedulers, including centralized, distributed, and hybrid schedulers, fail to ensure low latency for short jobs in large-scale and highly loaded systems. The key issues are the scalability in centralized schedulers, ineffective and inefficient probing and resource sharing in both distributed and hybrid schedulers.

Today's data center applications are predominantly data-intensive, calling for scaling out the workload to a large number of servers for parallel processing. Unfortunately, the existing scaling laws, notably, Amdahl's and Gustafson's laws are inadequate to characterize the scaling properties of dataintensive workloads. To fill this void, in this paper, we put forward a new scaling model, called In-Proportion and Scale-Out-induced scaling model (IPSO). IPSO generalizes the existing scaling models in two important aspects. First, it accounts for the possible in-proportion scaling, i.e., the scaling of the serial portion of the workload in proportion to the scaling of the parallelizable portion of the workload. Second, it takes into account the possible scaleout-induced scaling, i.e., the scaling of the collective overhead or workload induced by scaling out. IPSO exposes scaling properties of data-intensive workloads, rendering the existing scaling laws its special cases. In particular, IPSO reveals two new pathological scaling properties. Namely, the speedup may level off even in the case of the fixed-time workload underlying Gustafson's law, and it may peak and then fall as the system scales out. Extensive MapReduce and Spark-based case studies demonstrate that IPSO successfully captures diverse scaling properties of dataintensive applications. As a result, it can serve as a diagnostic tool to gain insights on or even uncover counter-intuitive root causes of observed scaling behaviors, especially pathological ones, for data-intensive applications. Finally, preliminary results also demonstrate the promising prospects of IPSO to facilitate effective resource provisioning to achieve the best speedup-versuscost tradeoffs for data-intensive applications.

With the widespread use and increasing popularity of cloud storage, more and more data are moved to the cloud, making cloud storage a platform for both data sharing among users, devices, and data backup for data reliability. Thus, it is critically important to ensure data consistency through efficient cloud synchronization (sync). The existing cloud synchronization schemes are either delta sync, which sends only the updated portion of a file but incurs high compute overhead of data deduplication for small files, or full sync, which avoids data deduplication by sending the full file but wastes network bandwidth and lengthens sync time by transferring significant amount of redundant data over the networks for large files. In this paper, we propose a hybrid cloud sync scheme, PandaSync, that combines full sync and delta sync dynamically based on file size and network conditions. To further improve small-file sync performance, we propose an optimization, Full2Sync, that merges the sync request with the file-sending request to reduce the number of network round-trips between the client and the cloud servers. The experiments conducted on our lightweight prototype implementation of PandaSync show that PandaSync reduces the sync time by an average of 85.1% and 74.6% from the delta sync scheme and full sync scheme, respectively.

The RΛ-Code is an efficient family of maximum distance separable (MDS) array codes of column distance 4, which involves two types of parity constraints: the row parity and the Λ parity formed by diagonal lines of slopes 1 and -1. Benefitting from the common expressions between the two parity constraints, the encoding and decoding complexities are distinctly lower than most (if not all) of other triple-erasure-correcting codes. It was left as an open problem generalizing the RΛ-Code to arbitrary column distances. In this paper, we present such a generalization, namely, we construct a family of MDS array codes being capable of correcting any prescribed number of erasures/errors by introducing multiple Λ parity constraints. Essentially, the generalized RΛ-Code is derived from a certain variant of the Blaum-Roth codes, and hence retains the error/erasure correcting capability of the latter. Compared with the Blaum-Roth codes, the generalized RΛ-Code has two advantages: a) by exploiting common expressions between row parity and different Λ parity constraints, and reusing the intermidate results during the syndrome calculations, it can encode and decode faster; and b) the memory footprint during encoding/decoding, and the I/O cost caused by degraded reads, are both reduced by 50%.

The workflows of the predominant user-facing datacenter services, including web searching and social networking, are underlaid by various Fork-Join structures. Due to the lack of understanding the performance of Fork-Join structures in general, today's datacenters often resort to resource overprovisioning, operating under low resource utilization, to meet stringent tail-latency service level objectives (SLOs) for such services. Hence, to achieve high resource utilization, while meeting stringent tail-latency SLOs, it is of paramount importance to be able to accurately predict the tail latency for a broad range of Fork-Join structures of practical interests.

This paper proposes an algorithm for optimal decentralized traffic engineering in communication networks. We aim at distributing the traffic among the available routes such that the network utility is maximized. In some practical applications, modeling network utility using non-concave functions is of particular interest, e.g., video streaming. Therefore, we tackle the problem of optimizing a generalized class of non-concave utility functions. The approach used to solve the resulting non-convex network utility maximization (NUM) problem relies on designing a sequence of convex relaxations whose solutions converge to that of the original problem. A distributed algorithm is proposed for the solution of the convex relaxation. Each user independently controls its traffic in a way that drives the overall network traffic allocation to an optimal operating point subject to network capacity constraints. All computations required by the algorithm are performed independently and locally at each user using local information and minimal communication overhead. The only non-local information needed is binary feedback from congested links. The robustness of the algorithm is demonstrated, where the traffic is shown to be automatically rerouted in case of a link failure or having new users joining the network. Numerical simulation results are presented to validate our findings.

With Inter-Cloud, distributed cloud and open cloud exchange (OCX) emerging, a comprehensive resource allocation approach is fundamental to highly competitive cloud market. Oriented to infrastructure as a service (IaaS), an intelligent economic approach for dynamic resource allocation (IEDA) is proposed with the improved combinatorial double auction protocol devised to enable various kinds of resources traded among multiple consumers and multiple providers at the same time enable task partitioning among multiple providers. To make bidding and asking reasonable in each round of the auction and determine eligible transaction relationship among providers and consumers, a price formation mechanism is proposed, which is consisted of a back propagation neural network (BPNN) based price prediction algorithm and a price matching algorithm. A reputation system is proposed and integrated to exclude dishonest participants from the cloud market. The winner determination problem (WDP) is solved by the improved paddy field algorithm (PFA). Simulation results have shown that IEDA can not only help maximize market surplus and surplus strength but also encourage participants to be honest.

Previous researches indicated that the performance of automatic prosodic boundary labeling benefited from syntactic phrase information for Mandarin. However, the influence of other syntactic features such as dependency has not been studied in-depth yet, especially on large scale corpus. This paper demonstrates the usefulness of rich syntactic features for Mandarin phrase boundary prediction. Both syntactic phrase and dependency features are considered in our methods. The experimental results show that rich syntactic features improve the performance of prosodic boundary prediction effectively. Index Terms: prosodic boundary, syntactic feature, syntactic phrase structure, dependency.

A key challenge to program a chip multiprocessor (CMP) is how to evaluate the performance of various possible program-task-to-core mapping choices during the initial programming phase, when the executable program is yet to be developed. In this paper, we put forward a thread-level modeling methodology to meet this challenge. The idea is to model thread-level activities only and overlook the instruction-level and microarchitectural details, except those having significant impact on the thread-level performance. Moreover, since the thread-level modeling is much coarser than the instruction-level modeling, the analysis at this level turns out to be significantly faster than that at the instruction level. These features make the methodology particularly amenable for fast performance evaluation of a large number of program-task-to-core mapping choices during the initial programming phase. Based on this methodology, an analytic modeling technique based on queuing theory and a fast simulation tool are developed, both allowing for fast performance prediction of CMPs. Case studies based on a large number of code samples available in IXP1200/2400 workbenches demonstrate that the maximal sustainable line rates estimated using our simulation tool and queuing network models are consistently within 6 and 8 percent of cycle-accurate simulation results, respectively.

With ever expanding design space and workload space in multicore era, it is a challenge to identify optimal design points quickly, desirable during the early stage of multicore processor design or programming phase. To meet this challenge, this paper proposes a theoretical framework that can capture the general performance properties for a class of multicore processors of interest over a large design space and workload space, free of scalability issues. The idea is to model multicore processors at the thread-level, overlooking instruction-level and micro architectural details. In particular, queuing network models that model multicore processors at the thread level are developed and solved based on an iterative procedure over a large design space and workload space. This framework scales to virtually unlimited numbers of cores and threads. The testing of the procedure demonstrates that the throughput performance for many-core processors with 1000 cores can be evaluated within a few seconds on an Intel Pentium 4 computer and the results are within 5% of the simulation data obtained based on a thread-level simulator.

This paper puts forward an end-to-end traffic control solution, which we refer to as TCP-Elastic Real-time SErvice (TERSE). TERSE provides a unified end-to-end traffic control protocol that enables Non-Real-time Elastic (NRE) service (i.e., the same as the one under the TCP control), Real-time Delay Adaptive (RDA) service, and Real-time Rate Adaptive (RRA) service. A specific service is enabled by properly setting a single parameter in the protocol. TERSE is underpinned by a sound design methodology. While having its roots in a well-known utility-based optimization approach, this methodology successfully addresses its limitations. It leads to a unified traffic control protocol, which has several provable properties, including fairness, convergence, and stability. The protocol is implemented in LINUX-based systems. The cross-Pacific testing of this protocol shows that it can achieve more than 1.2Mbps throughput performance, 150% higher than TCP-reno and 50% higher than TCP cubic. Both analytical and simulation studies also show that it can provide soft minimum-rate guarantees for both RRA and RDA traffic flows. Moreover, simulation also demonstrates that it is resilient to network resource shortage. As part of the protocol design, an effective utility function of TCP and the corresponding control law are derived, which captures TCP behavior not only in a qualitative but also in a quantitative manner.

Providing guaranteed end-to-end quality of service (QoS) is important for real-time applications such as Voice-over-IP. In this paper, a family of optimal, distributed, QoS- aware, end-to-end congestion control laws is derived. It enables a set of class of services (CoSs) including Assured Forwarding Service (AFS), Minimum Rate Guaranteed Service (MRGS), and Minimum Rate Guaranteed and Upper Bounded Rate Service (MRGUBS). These control laws maximize the same utility function as the TCP congestion control protocol does. As a result, they are by design TCP friendly. These control laws are implemented as window-based congestion control protocols, similar to the window-based TCP congestion control protocol. The performance of these protocols is tested based on ns-2 simulation. The results indicate that these protocols are indeed TCP friendly and can provide end-to-end service assurance as long as the percentage of network bandwidth consumed by the flows using these protocols is moderately small. Consequently, this family of QoS-aware control protocols has the potential to be used to provide end-to-end QoS guaranteed services for low- bandwidth applications, such as Voice-over-IP.

Easy management and maintenance and highly cost effective equipment make Ethernet an attractive technology for deploying metropolitan area networks (MANs). However, Ethernet has poor scalability due to use flat addressing scheme. To make metro-Ethernet more scalable, the provider edge (PE) nodes can use MAC-in-MAC (MiM) encapsulation scheme for frame forwarding. The MiM encapsulation scheme reduces the forwarding table size in the core nodes (CNs), but not in the PE nodes which need to maintain the entries of mapping end userpsilas MAC address to PEpsilas MAC address. In this paper, we propose an End user enabled MAC-in-MAC (EMiM) encapsulation scheme for Metro-Ethernet. In the proposed scheme, a userpsilas MAC address as well as its PE nodepsilas MAC address are associated with its address resolution protocol (ARP) entry. The modified ARP entry allows the end user to do MiM encapsulation. Hence a PE node does not need to maintain the entries of mapping end userpsilas MAC address to PE nodepsilas MAC address, thus significantly reducing the forwarding table size. The proposed scheme sustains Ethernetpsilas plug-and-play feature and provides high scalability. The simulation results show that the proposed scheme can reduce up to 65% maximum forwarding table size in PE nodes.

One of the most critical resource management issues in the use of ternary content addressable memory (TCAM) for packet classification/filtering is how to effectively support filtering rules with ranges, known as range matching. In this paper, a Dynamic Range Encoding Scheme (DRES) is proposed to significantly improve TCAM storage efficiency for range matching. Unlike the existing range encoding schemes requiring additional hardware support, DRES uses the TCAM coprocessor itself to assist range encoding. Hence, DRES can be readily programmed in a network processor using a TCAM coprocessor for packet classification. A salient feature of DRES is its ability to allow a subset of ranges to be encoded and hence to have full control over the range code size. This feature allows DRES to exploit the TCAM structure to maximize TCAM storage efficiency. DRES is a comprehensive solution, including a dynamic range selection algorithm, a search key encoding scheme, a range encoding scheme, and a dynamic encoded range update algorithm. While the dynamic range selection algorithm running in software allows optimal selection of ranges to be encoded to maximize the TCAM storage efficiency, the dynamic encoded range update algorithm allows the TCAM database to be updated lock-free without interrupting the TCAM database lookup process. DRES is evaluated based on real-world databases and the results show that DRES can reduce the TCAM storage expansion ratio from 6.20 to 1.23. The performance analysis of DRES based on a probabilistic model demonstrates that DRES significantly improves TCAM storage efficiency for a wide spectrum of range distributions.

A key issue in supporting quality of service (QoS) over wireless networks is to estimate the wireless channel capacity that varies randomly with time and space. In this paper, we propose a new model named effective channel capacity to predict the available channel capacity during any given time interval with the required degree of confidence. By leveraging large deviations techniques, we relate the fading channel capacity with the theory of effective bandwidth and establish a connection between the theory of effective bandwidth and information theory. We derive a set of algorithms and apply them to Nakagami-m fading channels. Consequently, our results provide a foundation for the performance analysis of upper layer algorithms and protocols for QoS provisioning over wireless networks

In this paper, we propose a new model named effective fade duration envelope to characterize the accumulative conditional fade durations of individual users or groups of users in wireless communication systems. The proposed model has the following novelties: (1) it introduces the statistical upper and lower bounds with the required degree of confidence for accumulative conditional fade durations during any given time interval; (2) it characterizes various conditional fading circumstances in wireless multi-user communication systems.

Caching, multithreading and the combination of them are the major latency hiding techniques adopted in network processors (NPs). Although they achieve great success in general purpose processors (GPPs), none of them have been well studied under the new context of packet processing. In this paper, we simulate the processing procedure of a four-PE (processing element) network processor and thoroughly evaluate different configurations of these techniques with real-life packet traces. Our major findings include: (1) In general, all of these latency hiding techniques effectively increase the traffic throughput and robustness of NP; but thread allocation policy has great impact on their performance. (2) If assigning packets of the same flow to different threads is allowed, multithreading keeps the PE in a working state as long as possible and less jitter in packet sending rate is resulted than caching schemes; otherwise, a cache with a reasonable size outperforms multithreading in almost all metrics such as traffic throughput, packet loss rate, queuing and total delay. (3) When access latency is comparable to the working time of execution unit, the performance of multithreading is more sensitive to packet arrival process and memory reference pattern than caching. In short, caching and multithreading have their respective advantages under different environment. In some cases, combined caching and multithreading tend to bring more performance gain than simply adding more threads or cache entries.

In this paper, we propose a theoretical framework to analyze the performance of upper layer algorithms and protocols such as scheduling disciplines which exploit the gain of multi-user diversity to optimize the utilization of wireless communication systems while supporting service differentiation between different users.

Packet classification has been a critical data path function for many emerging networking applications. An interesting approach is the use of ternary content addressable memory (TCAM) to achieve deterministic, high-speed packet classification performance. However, apart from high cost and power consumption, due to slow growing clock rate for memory technology, in general, the traditional single TCAM-based solution has difficulty to keep up with fast growing line rates. Moreover, the TCAM storage efficiency is largely affected by the need to support rules with ranges or range matching. In this paper, a distributed TCAM scheme that exploits chip-level-parallelism is proposed to greatly improve the throughput performance. This scheme seamlessly integrates with a range encoding scheme which not only solves the range matching problem, but also ensures a balanced high throughput performance. A thorough theoretical worst-case analysis of throughput, processing delay, and power consumption, as well as the experimental results show that the proposed solution can achieve scalable throughput performance matching up to OC768 line rate or higher. The added TCAM storage overhead is found to be reasonably small for the five real-world classifiers studied.

Packet classification (PC) has been a critical data path function for many emerging networking applications. An interesting approach is the use of TCAM to achieve deterministic, high speed PC However, apart from high cost and power consumption, due to slow growing clock rate for memory technology in general, PC based on the traditional single TCAM solution has difficulty to keep up with fast growing line rates. Moreover, the TCAM storage efficiency is largely affected by the need to support rules with ranges, or range matching. In this paper, a distributed TCAM scheme that exploits chip-level-parallelism is proposed to greatly improve the PC throughput. This scheme seamlessly integrates with a range encoding scheme, which not only solves the range matching problem but also ensures a balanced high throughput performance. Using commercially available TCAM chips, the proposed scheme achieves PC performance of more than 100 million packets per second (Mpps), matching OC768 (40 Gbps) line rate.

In this paper, fundamental conditions which bound the network processing unit (NPU) worst-case performance are established. In particular, these conditions formalize and integrate, with mathematical rigor, two existing approaches for finding the NPU performance bounds, i.e., the work conserving condition and instruction/latency budget based approaches. These fundamental conditions are then employed to derive tight memory access latency bounds for a data path flow with one memory access. Finally, one of these memory access latency bounds is successfully used to interpret a peculiar phenomenon found in Intel IXP1200, demonstrating the importance of analytical modeling for NPU performance analysis.

This work addresses the problem of optimal traffic engineering in a connectionless autonomous system. Based on nonlinear control theory, the approach taken in This work provides a family of optimal adaptation laws. These laws enable each node in the network to independently distribute traffic among any given set of next hops in an optimal way, as measured by a given global utility function of a general form. This optimal traffic distribution is achieved with minimum information exchange between neighboring nodes. Furthermore, this approach not only allows for optimal multiple forwarding paths but also enables multiple classes of service, e.g., classes of service defined in the differentiated services architecture. Moreover, the proposed decentralized control scheme enables optimal traffic redistribution in the case of link failures. Suboptimal control laws are also presented in an effort to reduce the computational burden imposed on the nodes of the network. Finally, an implementation of these laws with currently available technology is discussed.

Due to deterministic and fast lookup performance, ternary content addressable memory (TCAM) has recently been gaining popularity in general policy filtering (PF) for packet classification in high-speed networks. However, the PF table update poses significant challenges for efficient use of TCAM. To avoid erroneous and inconsistent rule matching, the traditional approach is to lock the PF table during the rule update period, but table locking has a negative impact on data path processing. In this paper, we propose a novel scheme, called Consistent Policy Table Update Algorithm (CoPTUA), for TCAM. Instead of minimizing the number of rule moves to reduce the locking time, CoPTUA maintains a consistent PF table throughout the update process, thus eliminating the need for locking the PF table while-ensuring correctness of rule matching. Our analysis and simulation show that, even for a PF table with 100,000 rules, an arbitrary number of rules can be updated simultaneously within 1 second in the worst case, provided that 2 percent of the PF table entries are empty. Thus, CoPTUA enforces any new rule in less than 1 second for practical PF table size with high memory utilization and without impacting data path processing.

In this paper, we address the problem of optimal decentralized traffic engineering when multiple paths are available for each call. More precisely, given a set of possible paths for each call, we aim at distributing the traffic among the available paths in order to maximize a given utility function. To solve this problem, we propose a large family of decentralized sending rate control laws having the property that each of the members of this family "steers" the traffic allocation to an optimal operation point. The approach taken relies on the control theory concept of Sliding Modes. These control laws allow each ingress node to independently adjust its traffic sending rates and/or redistribute its sending rates among multiple paths. The only nonlocal information needed is binary feedback from each congested node in the path. The control laws presented are applicable to a large class of utility functions, namely, utility functions that can be expressed as the sum of concave functions of the sending rates. We show that the technique can be applied not only to usual rate adaptive traffic with multiple paths, but also to rate adaptive traffic with minimum service requirements and/or maximum allowed sending rate and to assured service with targeted rate guarantee, all allowing for multiple paths. It is also shown that these control laws are robust with respect to failures; i.e., they automatically reroute traffic if a link failure occurs. Finally, we provide some insight on how to choose the "right" control law. In particular, we provide a way of choosing a member of the family of control laws that reduces the sending rate oscillation caused by implementation constraints like delays and quantization. An example of application of the approach delineated in this paper is also presented. This example provides some insights on the implementation aspects and illustrates the robustness of the control laws developed in this paper.

In the literature, there exit two types of cache consistency maintenance algorithms for mobile computing environments: stateless and stateful. In a stateless approach, the server is unaware of the cache contents at a mobile user (MU). Even though stateless approaches employ simple database management schemes, they lack scalability and ability to support user disconnectedness and mobility. On the other hand, a stateful approach is scalable for large database systems at the cost of nontrivial overhead due to server database management. We propose a novel algorithm, called Scalable Asynchronous Cache Consistency Scheme (SACCS), which inherits the positive features of both stateless and stateful approaches. SACCS provides a weak cache consistency for unreliable communication (e.g., wireless mobile) environments with small stale cache hit probability. It is also a highly scalable algorithm with minimum database management overhead. The properties are accomplished through the use of flag bits at the server cache (SC) and MU cache (MUC), an identifier (ID) in MUC for each entry after its invalidation, and estimated time-to-live (TTL) for each cached entry, as well as rendering of all valid entries of MUC to uncertain state when an MU wakes up. The stale cache hit probability is analyzed and also simulated under the Rayleigh fading model of error-prone wireless channels. Comprehensive simulation results show that the performance of SACCS is superior to those of other existing stateful and stateless algorithms in both single and multicell mobile environments.

Cell multiplexing is highly desirable to achieve increased transport efficiency in asynchronous transfer mode (ATM) pseudo wire emulation over a multiprotocol label switching (MPLS) network. However, supporting cell multiplexing can create performance bottleneck at the egress provider edge (PE) where fully programmable network processors are used to do the cell demultiplexing. In this paper, a scheme is proposed to allow a large number of cells to be demultiplexed, without having to implement special ASIC-based hardware optimized for cell demultiplexing. The idea is to configure concatenated data plane processing devices to share the cell demultiplexing task with the egress PE. Analysis shows that the use of just two processing devices is sufficient to achieve high transport efficiency. This may as well be achieved by the addition of one router without a central control card. Therefore, service providers can employ this scheme to enable cell multiplexing with limited add-on cost, while preserving the existing investment.

This paper aims at finding fundamental design principles for hierarchical Web caching. An analytical modeling technique is developed to characterize an uncooperative two-level hierarchical caching system where the least recently used (LRU) algorithm is locally run at each cache. With this modeling technique, we are able to identify a characteristic time for each cache, which plays a fundamental role in understanding the caching processes. In particular, a cache can be viewed roughly as a low-pass filter with its cutoff frequency equal to the inverse of the characteristic time. Documents with access frequencies lower than this cutoff frequency have good chances to pass through the cache without cache hits. This viewpoint enables us to take any branch of the cache tree as a tandem of low-pass filters at different cutoff frequencies, which further results in the finding of two fundamental design principles. Finally, to demonstrate how to use the principles to guide the caching algorithm design, we propose a cooperative hierarchical Web caching architecture based on these principles. Both model-based and real trace simulation studies show that the proposed cooperative architecture results in more than 50% memory saving and substantial central processing unit (CPU) power saving for the management and update of cache entries compared with the traditional uncooperative hierarchical caching architecture.

This paper studies the TCP performance over a DMT based ADSL network. The impact of DMT subchannel bit loading on the TCP throughput performance is studied. The simulation results show that there is a threshold for the signal-to-noise ratio (SNR) gap or bit error rate (BER) above which TCP throughput drops quickly. This threshold takes its value in a wide range depending on the TCP round-trip time as well as channel noise. This suggests that it would be insufficient to set a fixed target BER at, e.g., 10/sup -7/, when calculating the number of bits to be loaded in each subchannel. Instead, the bit loading should take TCP performance into account. Finally a dynamic bit loading scheme is proposed, which jointly optimizes the channel bit rate and TCP throughput performance.

This paper aims at finding fundamental design principles for hierarchical Web caching. An analytical modeling technique is developed to characterize an uncooperative two-level hierarchical caching system where the least recently used (LRU) algorithm is locally run at each cache. With this modeling technique, we are able to identify a characteristic time for each cache, which plays a fundamental role in understanding the caching processes. In particular, a cache can be viewed roughly as a lowpass filter with its cutoff frequency equal to the inverse of the characteristic time. Documents with access frequencies lower than this cutoff frequency will have good chances to pass through the cache without cache hits. This viewpoint enables us to take any branch of the cache tree as a tandem of lowpass filters at different cutoff frequencies, which further results in the finding of two fundamental design principles. Finally, to demonstrate how to use the principles to guide the caching algorithm design, we propose a cooperative hierarchical Web caching architecture based on these principles. The simulation study shows that the proposed cooperative architecture results in 50% saving of the cache resource compared with the traditional uncooperative hierarchical caching architecture.

As one of the most demanding applications in high speed networks, real-time services, such as audio and video, have stringent requirements on quality of service (QoS) guarantee, especially the bounded end-to-end delay and delay variations. Accordingly, the real-time services with deterministically bounded delay have been widely studied. The key is to design efficient scheduling algorithms. In this paper, a scheduling method, referred to as static-rotating-priority-queues (SRPQ), is proposed. Exact schedulability conditions for the method, which is important for call admission control, is also presented. The most desirable features of this method are its low complexity and reduced number of first-in-first-out (FIFO) queues in a scheduler while providing high bandwidth utilization.

In this paper, an analytical model is developed for the study of the impact of user behavior on the effectiveness of pricing, service quality, and link bandwidth allocation/sharing in a multiple class-of-service Internet. The model characterizes a variety of user behaviors, such as user sensitivity to service price/quality and user migration from one class of service to another. It also incorporates a usage-based pricing scheme and a general trunk reservation policy. In particular, a three-tier service model for a single link is considered. Numerical studies reveal the richness of this research issue and provide insights into the optimum selection of service price, quality and trunk reservation values with given user behaviors. We find that trunk reservation is effective in terms of service protection only when the call migration effect is small and the call migration effect can be harnessed by a proper design of service pricing/quality structure. We also find that user sensitivities have great impact on service price/quality structure design and care must be taken to avoid sudden network performance degradation at certain price/quality parameter values.

A challenge for the differentiated services (DS) architecture is how to simultaneously provide end-to-end assured service (AS) for transport control protocol (TCP) sessions and the best-effort service traffic, based on a single queue management. Our research results show that both intradomain and interdomain best-effort traffic can have an adverse impact on the interdomain TCP traffic. This paper proposes a technique to achieve the desired end-to-end throughput guarantee for TCP sessions. The proposed technique is composed of a series of measures which includes: (a) a path pinning mechanism for AS allowing aggregated bandwidth reservation for AS at each intermediate router in the forwarding path, (b) a packet marking strategy, (c) a dropping policy, (d) an adaptive dropping-threshold calculation method for queue management based on aggregated reserved bandwidth and real-time traffic measurement. The simulation results demonstrate that with this technique, a high end-to-end service assurance can be achieved for the TCP traffic, while a reasonably high throughput for best-effort traffic is maintained.

Next generation access routers and edge devices need to provide functionalities for layer-4 packet forwarding and firewall/security checks. Consequently, a challenging issue concerns how to achieve fast packet filtering and forwarding at low cost. This paper studies flow caching mechanisms for fast layer-4 packet forwarding. We show by model analysis that flow caching performance is not very sensitive to cache table lookup speed but it is sensitive to cache hit ratio. By making use of the available layer-4 information, we introduce two filtering modules to reduce the cache miss ratio. Using real trace simulation, we demonstrate that, by adding these two filtering modules, the cache miss can be decreased by up to 50% and the requirement for full header filtering speed has also been greatly reduced. The proposed flow caching mechanism is potentially useful for routers and switches where software based filtering modules are dynamically generated. There exists a widely installed base of routers with flow caching. The proposed mechanism provides a cost-effective migration path for upgrading these routers to value-added high speed routers with flow caching which offer integrated/differentiated services.

We propose a framework for the performance analysis and flow classification design of multi-protocol over ATM (MPOA) network. The study based on the real Internet/intranet traces shows that even at high cost with long delay for each shortcut setup, MPOA can offer significant performance gain over the traditional routed network in an inter ELAN communication environment. In comparison, the MPOA performance gain in an Internet backbone environment is much less significant, mainly because of the dominant short-lived flows contributed by both d.n.s. and h.t.t.p. applications. We also propose a flow classification algorithm, which substantially reduces the implementation complexity while achieves the same level of performance as compared to the default flow classification algorithm proposed by MPOA standard. A simple timeout mechanism is also introduced to the flow cache table management for significant performance improvement. We further develop a stable, adaptive flow classification algorithm, which achieves a near-optimal solution to minimize the constrained MPOA resource utilizations.

This paper proposes a novel approach for adaptive flow classification for flow-based hybrid switching systems to match the time varying traffic/resource characteristics. It minimizes the maximum of the system resource utilizations associated with packet processing power, signalling capacity and routing table size. After formulating the proposed flow adaptation as a min-max stochastic control problem, a heuristic algorithm is developed. The simulation study based on real traces shows the viability of the proposed flow adaptation for dynamic resource management in hybrid switching system design. The algorithm is simple to implement and only requires the adaptation of two global variables at time intervals of every few seconds based on the present usage of resources.

This paper develops fast algorithms for the construction of a circulant modulated rate process to match with the two primary traffic statistical functions: rate distribution f(x) and autocorrelation R(/spl tau/). Using existing modeling techniques, f(x) has to be limited to certain forms such as Gaussian or binomial; R(/spl tau/) can only consist of one or two exponential terms which are often real exponentials rather than complex. In reality, these two functions are collected from real traffic traces and generally expressed in a very complicated form. We only consider the traffic whose correlation function can be approximated by the sum of complex exponentials. Our emphasis is placed on the algorithm design for matching complicated R(/spl tau/) in traffic modeling. The typical CPU time for traffic modeling with R(/spl tau/) consisting of five or six complex exponential terms is found to be in the range of a few minutes by the proposed algorithms. Our study further shows an excellent agreement between the original traffic traces and the sequences generated by the matched analytical model. The selection of the measurement-window in traffic statistics collection for queueing performance analysis is also discussed.

This paper addresses a fundamental problem in resource management for flow-based hybrid switching systems. Such systems aim at efficient transport of layer-3 connectionless IP traffic over layer-2 connection-oriented ATM switching fabrics. One idea behind flow-based hybrid switching is to decompose individual IP packet streams into flows and then to classify them into short-lived and long-lived flows. While the short-lived flows are good for forwarding by the embedded software through permanent virtual connections (PVCs), the long-lived flows are more effectively transmitted by hardware through switched virtual connections (SVCs). Clearly the flow identification/classification mechanism will have great impact on the utilization of the system's resources. Our paper focuses on the resources which are directly associated with packet processing power, signaling capacity, and flow cache table size. Our study indicates that the presently available static flow classification methods have a vital shortcoming in balancing the utilization of the system's resources. We propose a novel approach for adaptive flow classification based on the min-max objective for the system resource utilizations to match with the time-varying traffic/resource characteristics based on the monotone properties and sensitivity analysis of the resource utilizations as functions of the control parameters, we first prove that the optimal solution of the static min-max problem is achieved at a unique balance point for the resource utilizations. With the intuition gained from the static results, we then design an adaptive controller formulated as a hierarchical stochastic automata control system with local search. The optimality of the proposed adaptive controller is tested against the static optimal control based on real trace simulations. The simulation studies in highly nonstationary environments show the viability of the proposed flow adaptation for dynamic resource management in hybrid switching system design. The algorithm is simple to implement and only requires the adaptation of two global variables at time intervals of every few seconds based on the present usage of resources.

This paper develops fast algorithms for construction of circulant modulated rate process to match with two primary traffic statistical functions: distribution f(x) and autocorrelation R(/spl tau/) of the rate process. Using existing modeling techniques, f(x) has to be limited to certain forms such as Gaussian or binomial; R(/spl tau/) can only consist of one or two exponential terms which are often real exponentials rather than complex. In reality, these two functions are collective for real traffic traces and generally expressed in a much more complicated form. Our emphasis here is placed on the algorithmic design for matching complicated R(/spl tau/) in traffic modeling. The typical CPU time for the traffic modeling with R(/spl tau/) consisting of five or six complex exponential terms is found in the range of a few minutes by the proposed algorithms. Our study further shows an excellent agreement between original traffic traces and sequences generated by the matched analytical model.