Configuration of overloaded servers with dynamic routing

We consider overload of servers in a network with dynamic routing of messages. The system consists of k servers and ℓ independent Poisson input flows. Messages from each flow are directed to m servers, and each message is directed to a server that is the least loaded at the moment of its arrival. In such a system, configuration of overloaded servers depends on the intensity of input flows. A similar effect was considered in [1] for a system with another geometry.     

Memory-aware kernel mechanism and policies for improving internode load balancing on NUMA systems

Although nonuniform memory access architecture provides better scalability for multicore systems, cores accessing memory on remote nodes take longer than those accessing on local nodes. Remote memory access accompanied by contention for internode interconnection degrades performance. Properly mapping threads to cores and data accessed to their nodes can substantially improve performance and energy efficiency. However, an operating system kernel's load-balancing activity may migrate threads across nodes, which thus messes up the thread mapping. Besides, subsequent data mapping behavior pays for the cost of page migration to reduce remote memory access. Once unsuitable threads are migrated, it is detrimental to system performance. This paper focuses on improving the kernel's internode load balancing on nonuniform memory access systems. We develop a memory-aware kernel mechanism and policies to reduce remote memory access incurred by internode thread migration. The Linux kernel's load balancing mechanism is modified to incorporate selection policies in the internode thread migration, and the kernel is modified to track the amount of memory used by each thread on each node. With this information, well-designed policies can then choose suitable threads for internode migration. The purpose is to avoid migrating a thread that might incur relatively more remote memory access and page migration. The experimental results show that with our mechanism and the proposed selection policies, the system performance is substantially increased when compared with the unmodified Linux kernel that does not consider memory usage and always migrates the first-fit thread in the runqueue that can be migrated to the target central processing unit.     

Dynamic non-preemptive re-allocation policies between two sites with reconfigurable servers

Dynamic server re-allocation can be very useful in real life computing applications. Since the load on many computing systems is not uniformly distributed to each server, it may be effective to transfer the less loaded servers to help the other more loaded ones. However, since transferring takes time, it may not be profitable to actually make the transfer. In this study we model this case with two queues. Each queue is served by one server which can be re-allocated, i.e. an operator may decide to switch a server to serve the other queue. The re-allocation policies we examine are non-preemptive, which implies that a server can be re-allocated if it is idle or has just served a customer. The model is studied with respect to the average cost criterion. We find the optimal re-allocation policy for various instances of the parameters. In addition, we provide a heuristic policy and use simulation experiments to compare it with the optimal one as well as the policy that uses no re-allocation at all.     

Dynamic cluster resource allocations for jobs with known andunknown memory demands

The cluster system we consider for load sharing is a compute farm which is a pool of networked server nodes providing high-performance computing for CPU-intensive, memory-intensive, and I/O active jobs in a batch mode. Existing resource management systems mainly target at balancing the usage of CPU loads among server nodes. With the rapid advancement of CPU chips, memory and disk access speed improvements significantly lag behind advancement of CPU speed, increasing the penalty for data movement, such as page faults and I/O operations, relative to normal CPU operations. Aiming at reducing the memory resource contention caused by page faults and I/O activities, we have developed and examined load sharing policies by considering effective usage of global memory in addition to CPU load balancing in clusters. We study two types of application workloads: 1) Memory demands are known in advance or are predictable and 2) memory demands are unknown and dynamically changed during execution. Besides using workload traces with known memory demands, we have also made kernel instrumentation to collect different types of workload execution traces to capture dynamic memory access patterns. Conducting different groups of trace-driven simulations, we show that our proposed policies can effectively improve overall job execution performance by well utilizing both CPU and memory resources with known and unknown memory demands     

Analysis of an open non-Markovian GI — (GI | ∞) K queueing network with high-rate renewal arrival process

We analyze an open non-Markovian queueing network with high-rate renewal arrival process, Markovian routing, arbitrary service policy, and unlimited number of servers at nodes. We obtain mean values for the number of busy servers at nodes of the queueing network in question. We show that, under an infinitely increasing arrival rate, the multivariate distribution of the number of busy servers at network nodes can be approximated by a multivariate normal distribution; we find parameters of this distribution.     

Traffic overflow in loss systems with selective trunk reservation

We consider two traffic streams competing for service at an n-server queuing system. Jobs from stream 1, the protected stream, are blocked only if all n servers are busy. Jobs from stream 2, the best effort stream, are blocked if n−r, r≥1, servers are busy. Blocked customers are diverted to a secondary group of c−n servers with, possibly, a different service rate. For the case r=1, we calculate the joint probabilities of the number of primary and secondary busy servers. For r>1, we describe a procedure for deriving the joint probabilities. These probabilities allow for the calculation of various performance measures including the overflow probabilities of the primary server and secondary server group. Our model is applicable to traffic control in communication networks that use the selective trunk reservation method.     

Using proxies to enhance TCP performance over hybrid fiber coaxial networks

Using cable modems that operate at several hundred times the speed of conventional telephone modems, many cable operators are beginning to offer World Wide Web access and other data services to residential subscribers. Initial experiences indicate that real-world hybrid fiber coaxial (HFC) networks are susceptible to a variety of radio-frequency impairments that significantly reduce the benefits of using high-speed cable modems. The effects of packet losses in the access network are particularly accentuated during subscriber accesses to remote servers on the Internet. The longer round-trip times in such accesses together with the high packet loss rate result in dramatic degradations in performance perceived by subscribers. This paper shows that by using proxy servers to handle all remote accesses from an HFC access network, the performance of remote accesses can be significantly enhanced even in cases when the proxy servers do not function as data caches. By handling packet losses that occur in the HFC network locally, at low latencies and without the remote server even being aware of the loss, a proxy server enables faster recovery from packet losses. Most importantly, since it controls data transmissions over the local HFC network, the proxy server's transmission control protocol (TCP) implementation can be optimized for the loss of characteristics of the HFC access network, enabling a significant increase in performance when the access network is lossy.     

Circle of interacting servers: Spontaneous collective behavior in the case of large fluctuations

We consider large fluctuations and overload of servers in a network with dynamic routing of messages. The servers form a circle. The number of input flows is equal to the number of servers; the messages of a flow are distributed between two neighboring servers; upon its arrival, a message is directed to the least loaded of these servers. Under the condition that at least two servers are overloaded, the number of overloaded servers in such a network depends on the rate of input flows. In particular, there exists a critical level of the input rate above which all servers are most probably overloaded.     

协同虚拟环境中的主动动态负载平衡算法

在基于多服务器体系结构的协同虚拟环境中,用户的随意运动引起某些服务器负载过大而导致系统性能的下降。主动动态负载平衡算法定义了3个主动触发条件,当服务器计算代价过大或任务划分不当引起服务器之间通信量过大或服务器负载接近饱和时,将兴趣隶属度大的用户转移给邻域服务器管理,保证了对虚拟环境中的用户代理进行合理划分。试验数据显示,该方法在均衡各个服务器负载的前提下,可以提高系统的实时性和服务质量。     

Analysis of a G/M/K/O queueing loss system with heterogeneous servers and retrials

Consider a G/M/K/O queueing loss system with K heterogeneous servers, exponentially distributed service times, no waiting room, a stationary counting arrival process, and an ordered entry. The ordered entry rule implies that, if the servers are indexed from 1 to K, units first arrive at the first server, then at the second server, and finally at the Kth server. In this queueing system, units that find the servers busy are not lost. Those units re-try to receive service by merging with the incoming units to be reconsidered for service by one of the free servers. This queueing system is analysed in terms of approximating the flows of units inside the system by a two parameter method. An example is introduced and approximation results are compared with those from a simulation study.     

Repositioning of distinguishable urban service units on networks

This paper examines the problem of repositioning mobile emergency service units on the urban Transportation network. Repositioning problems deal with real-time movements of available servers to better anticipate short-term future requests for service. It is assumed in the paper that q nodes of the network are designated as “home locations” for q distinguishable units. Depending on the status of other servers (busy or available), any particular available server can be moved to other locations (not necessarily home locations) in the network. Using Markovian Decision Theory, the policy space consists of decisions on where and when to move servers for any possible state. The paper includes an analysis of two cases based on the quality of information on the real-time location of non-stationary service units. In one case the assumption is that the dispatcher has perfect information whereas that in the other one it is assumed that no such information is available. The objective is to find the repositioning policy which minimizes the expected cost of operating the system in the long term.     

基于区块链的策略隐藏大数据访问控制方法

针对大数据应用中用户共享数据的访问控制由半可信云服务商实施所带来的隐私泄露、策略和访问日志易被篡改等问题, 提出一种基于区块链的策略隐藏大数据访问控制方法 (A policy-hidden big data access control method based on blockchain, PHAC). 该方法采用区块链技术实施访问控制以减少对服务商的信任依赖, 引入属性基加密(Attribute-based encryption, ABE)以及双线性映射技术, 实现在不泄露访问控制策略的前提下, 通过智能合约正确执行访问控制策略. 同时, 解耦访问控制策略, 简化用户策略的发布、更新和执行. 并应用链上和链下存储相结合方式, 解决智能合约和访问控制策略占用区块链节点资源不断增大的问题. 最后, 对该方法进行了理论分析和HyperLedger Fabric环境下的实验评估, 结果表明该方法能在策略隐藏情况下有效实现访问控制, 但不会给数据拥有者、区块链节点增加过多额外计算和存储开销.     

渲染器与Web服务器耦合实现远程体 渲染的交互优化

目的 为使用户随时随地获得需要较大计算和存储资源的交互真实感体渲染服务,设计并实现了一种面向Web的远程真实感体渲染方法。方法 计算量较大的实时体渲染任务由远端渲染服务器中的GPU加速完成并通过WebSocket将渲染图像发送至客户端;客户端浏览器只需负责接收显示渲染图像并监听发送用户交互事件。提出了一种输出系统耦合算法用以连接输出图像速率较大的渲染服务器和发送图像速率较慢的Web服务器。算法能根据Web服务器发送图像的情况动态调整每次图像输出的迭代计算次数,改变渲染服务器输出图像的时间间隔以达到与Web服务器发送速度相平衡,同时保持渲染服务器持续工作。结果 实验比较了在不同网络传输条件下,采用输出系统耦合算法与直接连接渲染器和Web服务器,渲染4个不同数据集所需的完成时间及帧率等性能评价指标。在局域网和广域网环境下,本文方法分别最多只需17 s和14 s即可完成整个渲染过程,而采用直接连接渲染器和Web服务器的方法则分别至少需要31 s和60 s才能完成整个渲染过程。实验结果表明采用输出系统耦合算法在不同网络条件下均可较大地缩短整个渲染过程所需时间,使用户在较短时间内获得高质量渲染图像。结论 本文渲染器与Web服务器耦合实现远程体渲染交互优化的方法可让用户使用与计算能力无关的桌面或移动设备通过网络透明使用高性能渲染系统;系统采用的输出系统耦合算法能够根据网络承载能力自适应调整渲染器输出速度,使用户在不同的网络环境中均可以较快的速度获得高质量渲染图像。     

A unified benchmarking and model-based framework for building QoS-aware streaming media services

A number of technology and workload trends motivate us to consider the appropriate resource allocation mechanisms and policies for streaming media services in shared cluster environments. We present MediaGuard – a model-based infrastructure for building streaming media services – that can efficiently determine the fraction of server resources required to support a particular client request over its expected lifetime. The proposed solution is based on a unified cost function that uses a single value to reflect overall resource requirements such as the CPU, disk, memory, and bandwidth necessary to support a particular media stream based on its bit rate and whether it is likely to be served from memory or disk. We design a novel, time-segment-based memory model of a media server to efficiently determine in linear time whether a request will incur memory or disk access when given the history of previous accesses and the behavior of the server's main memory file buffer cache. Using the MediaGuard framework, we design two media services: (1) an efficient and accurate admission control service for streaming media servers that accounts for the impact of the server's main memory file buffer cache, and (2) a shared streaming media hosting service that can efficiently allocate the predefined shares of server resources to the hosted media services, while providing performance isolation and QoS guarantees among the hosted services. Our evaluation shows that, relative to a pessimistic admission control policy that assumes that all content must be served from disk, MediaGuard (as well as services that are built using it) deliver a factor of two improvement in server throughput.     

移动自组网络高可用性服务的研究

Ad Hoe网络具有组网速度快、抗毁自愈能力强等优点，在军事和民用领域得到越来越广泛的应用．网络重构自恢复技术对于保证网络的可靠性、抗毁性与健壮性起着十分重要的作用．我们对Ad Hoe网络的重构自恢复进行研究，提出了一服务器组管理算法”．该算法采取主动冗余技术，通过增加、替换或删除服务器对服务器组进行维护和管理．以保证状态连续并减少服务中断，确保在提供服务的某些节点失效后整个系统仍然能够正常工作．仿真实验和数据分析表明，该算法提高了Ad Hoe网络的可用性，保证了网络的可靠性、抗毁性与健壮性．     

Markovian queueing systems with retrials and heterogeneous servers

The asymptotic performances of a random access and an ordered entry G/M/K/Oqueueing system with a stationary counting arrival process, K heterogeneous parallel servers, no waiting room and retrials are approximated based on a two-parameter method. In a random access system, units upon arrival are randomly assigned to one of the servers. In an ordered entry system, servers are indexed from 1 to K, and units first arrive at server i and if the server is found to be busy, those units arrive at server (i + 1), for i = 1 to K − 1. In both queueing systems, if units are not processed by one of the servers, those units are not lost, instead they retry to receive service by merging with the incoming arrival units.

To approximate the asymptotic performance of the above queueing systems, a recursive algorithm is suggested, and appropriate performance measures are presented to be used as comparison criteria at the design stage. Furthermore, numerical results are provided and approximation outcomes are compared against those from a simulation study.     

Study of a Controllable Queueing System with Unreliable Heterogeneous Servers

We consider a two-channel Markov queueing system with unreliable heterogeneous servers and a common queue. The claims are distributed among the servers with a threshold control policy. According to this policy, a server with the smaller average usage cost must be busy if the system itself is not empty, and the other server is used if the number of customers in the queue exceeds a certain threshold. We analyze the system in stationary mode. We present a method for computing the probabilities of system states and expressions for average performance and reliability characteristics. For the problem of minimizing average losses per unit of time, we obtain a heuristic formula that approximately computes the optimal threshold policy and proposes a method for computing the stationary distribution of the claim waiting time in the system.     

Tulip: A New Hash Based Cooperative Web Caching Architecture

With the exponential growth of WWW traffic, web proxy caching becomes a critical technique for Internet web services. Well-organized proxy caching systems with multiple servers can greatly reduce the user perceived latency and decrease the network bandwidth consumption. Thus, many research papers focused on improving web caching performance with the efficient coordination algorithms among multiple servers. Hash based algorithm is the most widely used server coordination mechanism, however, there's still a lot of technical issues need to be addressed. In this paper, we propose a new hash based web caching architecture, Tulip. Tulip aggregates web objects that are likely to be accessed together into object clusters and uses object clusters as the primary access units. Tulip extends the locality-based algorithm in UCFS to hash based web proxy systems and proposes a simple algorithm to reduce the data grouping overhead. It takes into consideration the access speed dispatch between memory and disk and replaces expensive small disk I/O with less large ones. In case a client request cannot be fulfilled by the server in the memory, the system fetches the whole cluster which contains the required object into memory, the future requests for other objects in the same cluster can be satisfied directly from memory and slow disk I/Os are avoided. It also introduces a simple and efficient data dupllication algorithm, few maintenance work need to be done in case of server join/leave or server failure. Along with the local caching strategy, Tulip achieves better fault tolerance and load balance capability with the minimal cost. Our simulation results show Tulip has better performance than previous approaches.     

Exploiting Regularities in Web Traffic Patterns for Cache Replacement

Abstract. Caching web pages at proxies and in web servers' memories can greatly enhance performance. Proxy caching is known to reduce network load and both proxy and server caching can significantly decrease latency. Web caching problems have different properties than traditional operating systems caching, and cache replacement can benefit by recognizing and exploiting these differences. We address two aspects of the predictability of traffic patterns: the overall load experienced by large proxy and web servers, and the distinct access patterns of individual pages. We formalize the notion of ``cache load' under various replacement policies, including LRU and LFU, and demonstrate that the trace of a large proxy server exhibits regular load. Predictable load allows for improved design, analysis, and experimental evaluation of replacement policies. We provide a simple and (near) optimal replacement policy when each page request has an associated distribution function on the next request time of the page. Without the predictable load assumption, no such online policy is possible and it is known that even obtaining an offline optimum is hard. For experiments, predictable load enables comparing and evaluating cache replacement policies using partial traces , containing requests made to only a subset of the pages. Our results are based on considering a simpler caching model which we call the interval caching model . We relate traditional and interval caching policies under predictable load, and derive (near)-optimal replacement policies from their optimal interval caching counterparts.     

Control of a Heterogeneous Two-Server Exponential Queueing System

A dynamic control policy known as "threshold queueing" is defined for scheduling customers from a Poisson source on a set of two exponential servers with dissimilar service rates. The slower server is invoked in response to instantaneous system loading as measured by the length of the queue of waiting customers. In a threshold queueing policy, a specific queue length is identified as a "threshold," beyond which the slower server is invoked. The slower server remains busy until it completes service on a customer and the queue length is less than its invocation threshold. Markov chain analysis is employed to analyze the performance of the threshold queueing policy and to develop optimality criteria. It is shown that probabilistic control is sub-optimal to minimize the mean number of customers in the system. An approximation to the optimum policy is analyzed which is computationally simple and suffices for most operational applications.