Showing papers on "Load balancing (computing) published in 2004"

Distributed clustering in ad-hoc sensor networks: a hybrid, energy-efficient approach

Abstract: Prolonged network lifetime, scalability, and load balancing are important requirements for many ad-hoc sensor network applications. Clustering sensor nodes is an effective technique for achieving these goals. In this work, we propose a new energy-efficient approach for clustering nodes in ad-hoc sensor networks. Based on this approach, we present a protocol, HEED (hybrid energy-efficient distributed clustering), that periodically selects cluster heads according to a hybrid of their residual energy and a secondary parameter, such as node proximity to its neighbors or node degree. HEED does not make any assumptions about the distribution or density of nodes, or about node capabilities, e.g., location-awareness. The clustering process terminates in O(1) iterations, and does not depend on the network topology or size. The protocol incurs low overhead in terms of processing cycles and messages exchanged. It also achieves fairly uniform cluster head distribution across the network. A careful selection of the secondary clustering parameter can balance load among cluster heads. Our simulation results demonstrate that HEED outperforms weight-based clustering protocols in terms of several cluster characteristics. We also apply our approach to a simple application to demonstrate its effectiveness in prolonging the network lifetime and supporting data aggregation.

Deep packet inspection using parallel bloom filters

Abstract: There is a class of packet processing applications that inspect packets deeper than the protocol headers to analyze content. For instance, network security applications must drop packets containing certain malicious Internet worms or computer viruses carried in a packet payload. Content forwarding applications look at the hypertext transport protocol headers and distribute the requests among the servers for load balancing. Packet inspection applications, when deployed at router ports, must operate at wire speeds. With networking speeds doubling every year, it is becoming increasingly difficult for software-based packet monitors to keep up with the line rates. We describe a hardware-based technique using Bloom filters, which can detect strings in streaming data without degrading network throughput. A Bloom filter is a data structure that stores a set of signatures compactly by computing multiple hash functions on each member of the set. This technique queries a database of strings to check for the membership of a particular string. The answer to this query can be false positive but never a false negative. An important property of this data structure is that the computation time involved in performing the query is independent of the number of strings in the database provided the memory used by the data structure scales linearly with the number of strings stored in it. Furthermore, the amount of storage required by the Bloom filter for each string is independent of its length.

Mercury: supporting scalable multi-attribute range queries

Abstract: This paper presents the design of Mercury, a scalable protocol for supporting multi-attribute range-based searches. Mercury differs from previous range-based query systems in that it supports multiple attributes as well as performs explicit load balancing. To guarantee efficient routing and load balancing, Mercury uses novel light-weight sampling mechanisms for uniformly sampling random nodes in a highly dynamic overlay network. Our evaluation shows that Mercury is able to achieve its goals of logarithmic-hop routing and near-uniform load balancing.We also show that Mercury can be used to solve a key problem for an important class of distributed applications: distributed state maintenance for distributed games. We show that the Mercury-based solution is easy to use, and that it reduces the game's messaging overheard significantly compared to a naive approach.

Fairness and load balancing in wireless LANs using association control

Abstract: Recent studies on operational wireless LANs (WLANs) have shown that user load is often unevenly distributed among wireless access points (APs). This unbalanced load results in unfair bandwidth allocation among users. We observe that the unbalanced load and unfair bandwidth allocation can be greatly alleviated by intelligently associating users to APs, termed association control, rather than having users greedily associate APs of best received signal strength.In this study, we present an efficient algorithmic solution to determine the user-AP associations that ensure max-min fair bandwidth allocation. We provide a rigorous formulation of the association control problem that considers bandwidth constraints of both the wireless and backhaul links. Our formulation indicates the strong correlation between fairness and load balancing, which enables us to use load balancing techniques for obtaining near optimal max-min fair bandwidth allocation. Since this problem is NP-hard, we present algorithms that achieve a constant-factor approximate max-min fair bandwidth allocation. First, we calculate a fractional load balancing solution, where users can be associated with multiple APs simultaneously. This solution guarantees the fairest bandwidth allocation in terms of max-min fairness. Then, by utilizing a rounding method we obtain an efficient integral association. In particular, we provide a 2-approximation algorithm for unweighted greedy users and a 3-approximation algorithm for weighted and bounded-demand users. In addition to bandwidth fairness, we also consider time fairness and we show it can be solved optimally. We further extend our schemes for the on-line case where users may join and leave. Our simulations demonstrate that the proposed algorithms achieve close to optimal load balancing and max-min fairness and they outperform commonly used heuristic approaches.

Load balancing in dynamic structured P2P systems

Abstract: Most P2P systems that provide a DHT abstraction distribute objects randomly among "peer nodes" in a way that results in some nodes having /spl theta/(log N) times as many objects as the average node. Further imbalance may result due to non-uniform distribution of objects in the identifier space and a high degree of heterogeneity in object loads and node capacities. Additionally, a node's load may vary greatly over time since the system can be expected to experience continuous insertions and deletions of objects, skewed object arrival patterns, and continuous arrival and departure of nodes. We propose an algorithm for load balancing in such heterogeneous, dynamic P2P systems. Our simulation results show that in the face of rapid arrivals and departures of objects of widely varying load, our algorithm achieves load balancing for system utilizations as high as 90% while moving only about 8% of the load that arrives into the system. Similarly, in a dynamic system where nodes arrive and depart, our algorithm moves less than 60% of the load the underlying DHT moves due to node arrivals and departures. Finally, we show that our distributed algorithm performs only negligibly worse than a similar centralized algorithm, and that node heterogeneity helps, not hurts, the scalability of our algorithm.

Simple efficient load balancing algorithms for peer-to-peer systems

Abstract: Load balancing is a critical issue for the efficient operation of peer-to-peer networks. We give two new load-balancing protocols whose provable performance guarantees are within a constant factor of optimal. Our protocols refine the consistent hashing data structure that underlies the Chord (and Koorde) P2P network. Both preserve Chord's logarithmic query time and near-optimal data migration cost.Consistent hashing is an instance of the distributed hash table (DHT) paradigm for assigning items to nodes in a peer-to-peer system: items and nodes are mapped to a common address space, and nodes have to store all items residing closeby in the address space.Our first protocol balances the distribution of the key address space to nodes, which yields a load-balanced system when the DHT maps items "randomly" into the address space. To our knowledge, this yields the first P2P scheme simultaneously achieving O(log n) degree, O(log n) look-up cost, and constant-factor load balance (previous schemes settled for any two of the three).Our second protocol aims to directly balance the distribution of items among the nodes. This is useful when the distribution of items in the address space cannot be randomized. We give a simple protocol that balances load by moving nodes to arbitrary locations "where they are needed." As an application, we use the last protocol to give an optimal implementation of a distributed data structure for range searches on ordered data.

Network load balancing with host status information

Abstract: In a first exemplary media implementation, one or more processor-accessible media include processor-executable instructions that, when executed, direct a system to perform actions that include: accumulating host status information at multiple hosts; and sending the accumulated host status information from the multiple hosts In a second exemplary media implementation, one or more processor-accessible media include processor-executable instructions that, when executed, direct a system to perform actions that include: receiving host status information from multiple hosts; and making load balancing decisions responsive to the received host status information In a third exemplary media implementation, one or more processor-accessible media include processor-executable instructions that, when executed, direct a system to perform actions that include: determining health and load information on a per application basis; and selecting an application from among multiple applications responsive to the health and load information

Adaptive load balancing in a multi-processor graphics processing system

Abstract: Systems and methods for balancing a load among multiple graphics processors that render different portions of a frame A display area is partitioned into portions for each of two (or more) graphics processors The graphics processors render their respective portions of a frame and return feedback data indicating completion of the rendering Based on the feedback data, an imbalance can be detected between respective loads of two of the graphics processors In the event that an imbalance exists, the display area is re-partitioned to increase a size of the portion assigned to the less heavily loaded processor and to decrease a size of the portion assigned to the more heavily loaded processor

Load balancing in ad hoc networks: single-path routing vs. multi-path routing

Abstract: Multi-path routing has been studied thoroughly in the context of wired networks. Ii has been shown that using multiple paths to route messages between any source-destination pair of nodes (instead of using a single path) balances the load more evenly throughout the network. The common belief is that the same is true for ad hoc networks, i.e., multi-path routing balances the load significantly better than single-path routing. We show that this is not necessarily the case. We introduce a new model for evaluating the load balance under multi-path routing, when the paths chosen are the first K shortest paths (for a pre-specified K). Using this model, we show that unless we use a very large number of paths (which is very costly and therefore infeasible) the load distribution is almost the same as single shortest path routing. This is in contrary to the previous existing results which assume that multi-path routing distributes the load uniformly.

Load balancing service

Abstract: A load balancing service for a plurality of customers performs load balancing among a plurality of customer Web servers. Requests for Web content are load balanced across the customer Web servers. The load balancing service provider charges a fee to the customers for the load balancing service. A caching service is also provided that comprises a plurality of caching servers connected to a network. The caching servers host customer content that can be cached and stored, e.g., images, video, text, and/or software. The caching servers respond to requests for Web content from clients. The load balancing service provider charges a fee to the customers for the Web caching service.

An algebraic approach to practical and scalable overlay network monitoring

Abstract: Overlay network monitoring enables distributed Internet applications to detect and recover from path outages and periods of degraded performance within seconds. For an overlay network with n end hosts, existing systems either require O(n2) measurements, and thus lack scalability, or can only estimate the latency but not congestion or failures. Our earlier extended abstract [1] briefly proposes an algebraic approach that selectively monitors k linearly independent paths that can fully describe all the O(n2) paths. The loss rates and latency of these k paths can be used to estimate the loss rates and latency of all other paths. Our scheme only assumes knowledge of the underlying IP topology, with links dynamically varying between lossy and normal.In this paper, we improve, implement and extensively evaluate such a monitoring system. We further make the following contributions: i) scalability analysis indicating that for reasonably large n (e.g., 100), the growth of k is bounded as O(n log n), ii) efficient adaptation algorithms for topology changes, such as the addition or removal of end hosts and routing changes, iii) measurement load balancing schemes, and iv) topology measurement error handling. Both simulation and Internet experiments demonstrate we obtain highly accurate path loss rate estimation while adapting to topology changes within seconds and handling topology errors.

Load balancing in overlapping wireless LAN cells

Abstract: We propose a load-balancing scheme for overlapping wireless LAN cells. Agents running in each access point broadcast periodically the local load level via the Ethernet backbone and determines whether the access point is overloaded, balanced or under-loaded by comparing it with the received reports. The load metric is the access point throughput. Overloaded access points force the handoff of some stations to balance the load. Only the under-loaded access points accept the roaming stations in minimizing the number of handoffs. We show via experimental evaluation that our balancing scheme increases the total wireless network throughput and decreases the cell delay.

Meta algorithms for hierarchical Web caches

Abstract: Large scale hierarchical caches for Web content have been deployed widely in an attempt to reduce delivery delays and bandwidth consumption and also to improve the scalability of content dissemination through the World Wide Web. Irrespective of the specific replacement algorithm employed in each cache, a de facto characteristic of contemporary hierarchical caches is that a hit for a document at an l-level cache leads to the caching of the document in all intermediate caches (levels l-1,..., 1) on the path towards the leaf cache that received the initial request. This paper presents various algorithms that revises this standard behavior and attempts to be more selective in choosing the caches that gets to store a local copy of the requested document. As these algorithms operate independently of the actual replacement algorithm running in each individual cache, they are referred to as meta algorithms. Three new meta algorithms are proposed and compared against the de facto one and a recently proposed one by means of synthetic and trace-driven simulations. The best of the new meta algorithms appears to be leading to improved performance under most simulated scenarios, especially under a low availability of storage. The latter observation makes the presented meta algorithms particularly favorable for the handling of large data objects such as stored music files or short video clips. Additionally, a simple load balancing algorithm that is based on the concept of meta algorithms is proposed and evaluated. The algorithm is shown to be able to provide for an effective balancing of load thus possibly addressing the recently discovered "filtering-effect" in hierarchical Web caches.

A load balancing method for a wireless area network

Abstract: The present invention relates to a load balancing method for a wireless local area network, which includes the following steps: initializing the load balancing group, self-organizing to select an online access point from the load balancing group as a load balancing agent center, each access point of the said balancing group managing and processing the load balance in accordance with the load level and balance optimizing policy, therefore implements the network load balance. The method of the invention improves reliability of the network load information, presents the load information from invalidly broadcasting in the wireless local area network, therefore reduces the additional overhead for managing the load balance of the wireless network, particularly in the network that having three layers switching equipment, self-organizing to select the agent center can utilize the network resource in effect, therefore managing the dynamic load balance information much more effectively, reducing the dependence on the upper server, and managing the network much more flexibly.

Adaptive replication in peer-to-peer systems

Abstract: Peer-to-peer systems can be used to form a low-latency decentralized data delivery system. Structured peer-to-peer systems provide both low latency and excellent load balance with uniform query and data distributions. Under the more common skewed access distributions, however, individual nodes are easily overloaded, resulting in poor global performance and lost messages. This paper describes a lightweight, adaptive, and system-neutral replication protocol, called LAR, that maintains low access latencies and good load balance even under highly skewed demand. We apply LAR to Chord and show that it has lower overhead and better performance than existing replication strategies.

Towards scalable multiprocessor virtual machines

Abstract: A multiprocessor virtual machine benefits its guest operating system in supporting scalable job throughput and request latency-useful properties in server consolidation where servers require several of the system processors for steady state or to handle load bursts. Typical operating systems, optimized for multiprocessor systems in their use of spin-locks for critical sections, can defeat flexible virtual machine scheduling due to lock-holder preemption and misbalanced load. The virtual machine must assist the guest operating system to avoid lock-holder preemption and to schedule jobs with knowledge of asymmetric processor allocation. We want to support a virtual machine environment with flexible scheduling policies, while maximizing guest performance. This paper presents solutions to avoid lock-holder preemption for both fully virtualized and paravirtualized environments. Experiments show that we can nearly eliminate the effects of lock-holder preemption. Furthermore, the paper presents a scheduler feedback mechanism that despite the presence of asymmetric processor allocation achieves optimal and fair load balancing in the guest operating system.

Simple efficient load balancing algorithms for peer-to-peer systems

Abstract: Load balancing is a critical issue for the efficient operation of peer-to-peer networks. We give two new load-balancing protocols whose provable performance guarantees are within a constant factor of optimal. Our protocols refine the consistent hashing data structure that underlies the Chord (and Koorde) P2P network. Both preserve Chord’s logarithmic query time and near-optimal data migration cost. Our first protocol balances the distribution of the key address space to nodes, which yields a load-balanced system when the DHT maps items “randomly” into the address space. To our knowledge, this yields the first P2P scheme simultaneously achieving O(log n) degree, O(log n) look-up cost, and constant-factor load balance (previous schemes settled for any two of the three). Our second protocol aims to directly balance the distribution of items among the nodes. This is useful when the distribution of items in the address space cannot be randomized – for example, if we wish to support range-searches on “ordered” keys. We give a simple protocol that balances load by moving nodes to arbitrary locations “where they are needed.” As an application, we use the last protocol to give an optimal implementation of a distributed data structure for range searches on ordered data.

Method and apparatus for facilitating fulfillment of web-service requests on a communication network

Abstract: Fulfillment of web-service requests may be facilitated by intelligently load balancing the web-service requests between servers or server clusters configured to perform the requested web-service. Load balancing may be based on the type of request, target class of server, whether the request is likely to spawn any subsequent requests, relevant historical information, other requests, current and anticipated work load on the servers, the current ability of the servers to handle additional requests, the numbers type and schedule of requests in a queue waiting to be allocated to one or more of the servers, and numerous other factors that may affect the servers' ability to process the request. Requests may be classified to enable historical correlation between how servers have handled previous requests and the present request. Additionally, requests may be scheduled for future execution and monitored during execution.

Semantic small world: an overlay network for peer-to-peer search

Abstract: For a peer-to-peer (P2P) system holding massive amount of data, efficient semantic based search for resources (such as data or services) is a key determinant to its scalability. This work presents the design of an overlay network, namely semantic small world (SSW), that facilitates efficient semantic based search in P2P systems. SSW is based on three innovative ideas: 1) small world network; 2) semantic clustering; 3) dimension reduction. Peers in SSW are clustered according to the semantics of their local data and self-organized as a small world overlay network. To address the maintenance issue of high dimensional overlay networks, a dynamic dimension reduction method, called adaptive space linearization, is used to construct a one-dimensional SSW that supports operations in the high dimensional semantic space. SSW achieves a very competitive trade-off between the search latencies/traffic and maintenance overheads. Through extensive simulations, we show that SSW is much more scalable to very large network sizes and very large numbers of data objects compared to pSearch, the state-of-the-art semantic-based search technique for P2P systems. In addition, SSW is adaptive to distribution of data and locality of interest; is very resilient to failures; and has good load balancing property.

On using peer-to-peer communication in cellular wireless data networks

Abstract: A recent class of approaches for enhancing the performance of cellular wireless data networks has focused on improving the underlying network model. It has been shown that using the peer-to-peer network model, a mode of communication typically seen in ad hoc wireless networks, can result in performance improvements such as increased data rate, reduced transmission power, better load balancing, and enhanced network coverage. However, the true impact of adopting the peer-to-peer network model in such an environment is yet to be fully understood. In this paper, we investigate the performance benefits and drawbacks of using the peer-to-peer network model for Internet access in cellular wireless data networks. We find that, although the peer-to-peer network model has significantly better spatial reuse characteristics, the improved spatial reuse does not translate into better throughput performance. Instead, we observe that using the peer-to-peer network model as-is might actually degrade the throughput performance of the network. We identify and discuss the reasons behind these observations. Using the insights gained through the performance evaluations, we then propose two categories of approaches to improve the performance of the peer-to-peer network model: approaches that leverage assistance from the base station and approaches that leverage the relaying capability of multihomed hosts. Through simulation results, we show that using the peer-to-peer network model in cellular wireless data networks is a promising approach when the network model is complemented with appropriate mechanisms.

Efficient and robust query processing in dynamic environments using random walk techniques

Abstract: Many existing systems for sensor networks rely on state information stored in the nodes for proper operation (e.g., pointers to parent in a spanning tree, routing information, etc). In dynamic environments, such systems must adopt failure recovery mechanisms, which significantly increase the complexity and impact the overall performance. In this paper, we investigate alternative schemes for query processing based on random walk techniques. The robustness of this approach under dynamics follows from the simplicity of the process, which only requires the connectivity of the neighborhood to keep moving. In addition we show that visiting a constant fraction of sensor network, say 80%, using a random walk is efficient in number of messages and sufficient for answering many interesting queries with high quality. Finally, the natural behavior of a random walk, also provide the important properties of load-balancing and scalability.

Virtual clustering and load balancing servers

Abstract: A method and system is provided intelligent network dispatchers that are capable of routing requests to one or more servers (e.g., LDAP servers) based on the requested application or the identity of the requester. The servers may be grouped into virtual clusters so that a particular cluster may service requests associated with a particular application or based on a type of customer (e.g., preferred or general customer) in order to grade levels of service. Also, a system resource monitor may provide continual monitoring of servers in virtual clusters to determine when one or more servers are overburdened or under burdened based on statistical performance data. As a servers performance changes, the system resource monitor may advise the intelligent network dispatcher causing a change in the rate of requests to the overburdened/under burdened server. The server may be temporarily removed from the virtual cluster until performance levels normalize. Further, servers may be re-assigned to other virtual clusters on demand to provide increased capacity to a virtual cluster.

System and methods of cooperatively load-balancing clustered servers

Abstract: Host computer systems (42) dynamically engage in independent transactions with servers of a server cluster to request performance of a network service, preferably a policy-based transfer processing of data. The host computer systems operate from an identification of the servers in the cluster to autonomously select servers for transactions qualified on server performance information gathered in prior transactions. Server performance information may include load and weight values that reflect the performance status of the selected server and a server localized policy evaluation of service request attribute information provided in conjunction with the service request. The load selection of specific servers for individual transactions is balanced implicitly through the cooperation of the host computer systems (42) and servers of the server cluster.

Coordinated load balancing, handoff/cell-site selection, and scheduling in multi-cell packet data systems

Abstract: We investigate a wireless system of multiple cells, each having a downlink shared channel in support of high-speed packet data services. In practice, such a system consists of hierarchically organized entities including a central server, Base Stations (BSs), and Mobile Stations (MSs). Our goal is to improve global resource utilization and reduce regional congestion given asymmetric arrivals and departures of mobile users. For this purpose, we propose a scalable cross-layer framework to coordinate packet-level scheduling, call-level cell-site selection and handoff, and system-level loading balancing based on load, throughput, and channel measurements at different layers. In this framework, an opportunistic scheduling algorithm---the weighted Alpha-Rule---exploits multiuser diversity gain in each cell independently, trading aggregate (mean) downlink throughput for fairness and minimum rate guarantees among MSs. Each MS adapts to its channel dynamics and the load fluctuations in neighboring cells, in accordance with MSs' mobility and their arrivals or departures, by initiating load-aware handoff and cell-site selection. The central server adjusts the scheduling parameters of each cell to coordinate cells' coverage, or cell breathing, by prompting distributed MS handoffs. Across the whole system, BSs and MSs constantly monitor their load, throughput, or channel quality in order to facilitate the overall system coordination.Our specific contributions in such a framework are highlighted by the minimum-rate guaranteed Weighted Alpha-Rule scheduling, the load-aware MS handoff/cell-site selection, and the Media Access Control (MAC)-layer cell breathing. Our evaluations show that the proposed framework can improve the global resource utilization and load balancing, which translates into a smaller blocking rate of MS arrivals without extra resources, while the aggregate throughput remains roughly the same or improved around the hot-spots. Our tests also show that the coordinated system is robust to dynamic load fluctuations and is scalable to both system size and MS population.

System and method for load balancing virtual machines in a computer network

Abstract: A system and method for balancing the load of virtual machines among the physical servers of a network is disclosed. When a server of the network is identified as exceeding its total resource utilization, a virtual machine within the server is selected for migration to another physical server within the network. A virtual machine may be selected for migration on the basis of an identification of the virtual machine that has the lowest level of resource requirements. The target physical server may be selected on the basis of an identification of a server that has the highest level of resource availability.

Arrangement in a server for providing dynamic domain name system services for each received request

Abstract: A Domain Name System (DNS) server is configured for dynamically selecting, for a client device, a selected resolution from available resolutions, the selected resolution identifying at least one destination for the specified service. The dynamic selection of the selected resolution is based on an attribute of the client device and/or a determined attribute of the network. Hence, the selected resolution directs the client device to a specific server based on prescribed selection criteria, for example service level agreements, the location of the client device, network performance or detected congestion conditions, authentication of the user of the client device, etc. The selected resolution also can be for enforcement of load balancing policies.

System and method for failure recovery and load balancing in a cluster network

Abstract: A system and method for failure recovery in a cluster network is disclosed in which each application of each node of the cluster network is assigned a preferred failover node. The dynamic selection of a preferred failover node for each application is made on the basis of the processor and memory requirements of the application and the processor and memory usage of each node of the cluster network.

Analysis of wireless information locality and association patterns in a campus

Abstract: Our goal is to explore characteristics of the environment that provide opportunities for caching, prefetching, coverage planning, and resource reservation. We conduct a one-month measurement study of locality phenomena among wireless Web users and their association patterns on a major university campus using the IEEE 802.11 wireless infrastructure. We evaluate the performance of different caching paradigms, such as single user cache, cache attached to an access point (AP), and peer-to-peer caching. In several settings such caching mechanisms could be beneficial. Unlike other measurement studies in wired networks in which 25% to 40% of documents draw 70% of Web access, our traces indicate that 13% of unique URLS draws this number of Web accesses. In addition, the overall ideal hit ratio of the user cache, cache attached to an access point, and peer-to-peer caching paradigms (where peers are coresident within an AP) are 51%, 55%, and 23%, respectively. We distinguish wireless clients based on their inter-building mobility, their visits to APs, their continuous walks in the wireless infrastructure, and their wireless information access during these periods. We model the associations as a Markov chain using as state information the most recent AP visits. We can predict with high probability (86%) the next AP with which a wireless client will associate. Also, there are APs with a high percentage of user revisits. Such measurements can benefit protocols and algorithms that aim to improve the performance of the wireless infrastructures by load balancing, admission control, and resource reservation across APs.

Host-level policies for global server load balancing

Abstract: In a network, a user can configure host-level policies usable for load balancing traffic to servers of a domain. A global server load balancing (GSLB) switch provides load balancing to the servers, and is configured with the GSLB host-level policies. Users can define a host-level policy (alternatively or additionally to a globally applied GSLB policy) and apply the host-level policy to hosts in domains configured on the GSLB switch. Thus, the user can enable different policies for different hosts. This allows the user to have the flexibility to control metrics used for selection of a best address for querying clients, as well as the metric order and additional parameters used in the GSLB process, at the host level.