Distributed Load Balancing Algorithm Research Articles

Reinforcement learning-based dynamic load balancing in edge computing networks

Edge computing (EC) has emerged as a paradigm aimed at reducing data transmission latency by bringing computing resources closer to users. However, the limited scale and constrained processing power of EC pose challenges in matching the resource availability of larger cloud networks. Load balancing (LB) algorithms play a crucial role in distributing workload among edge servers and minimizing user latency. This paper presents a novel set of distributed LB algorithms that leverage machine learning techniques to overcome the three limitations of our previous LB algorithm, EVBLB: (i) its reliance on static time intervals for execution, (ii) the need for comprehensive information about all server resources and queued requests for neighbor selection, and (iii) the use of a central coordinator to dispatch incoming user requests over edge servers. To offer increased control, custom configuration, and scalability for LB on edge servers, we propose three efficient algorithms: Q-learning (QL), multi-armed bandit (MAB), and gradient bandit (GB) algorithms. The QL algorithm predicts the subsequent execution time of the EVBLB algorithm by incorporating rewards obtained from previous executions, thereby improving performance across various metrics. The MAB and GB algorithms prioritize near-optimal neighbor node servers while considering dynamic changes in request rate, request size, and edge server resources. Through simulations, we evaluate and compare the algorithms in terms of network throughput, average user response time, and a novel LB metric for workload distribution across edge servers.

Distributed Multi-Agent Deep Q-Learning for Load Balancing User Association in Dense Networks

Distributed learning can lead to effective user association with low overhead, but faces significant challenges in incorporating load balancing at all base stations (BS) because of coupling constraints. In this paper, we propose a distributed multi-agent deep Q-learning model for user association to satisfy the load balancing constraint at every BS. In particular, we design a deep Q-network (DQN) with target Q-network and experience replay buffer at each user as an agent, and propose a multi-agent matching policy to control the number of users connected to each BS for load balancing. The policy enhances network throughput performance by implementing a novel updating rule for the preference list at each BS. The proposed multi-agent DQN model operates in a fully distributed manner where each agent only uses local information and requires no information exchange between agents. Simulation results demonstrate that our proposed algorithm outperforms a conventional distributed load balancing algorithm and approaches a centralized scheme performance, while exhibiting fast convergence and high adaptability to channel changes.

Distributed load balancing algorithm considering QoS for next generation multi-RAT HetNets

In this paper, we propose a distributed load balancing algorithm for multi-radio access technology (multi-RAT) heterogeneous networks (HetNets). The random deployment of small cells and the user mobility make the network load distribution uneven, which degrades overall network capacity and the QoS for users. Furthermore, the disparate capabilities of multiple RATs such as the different propagation delays in terrestrial and non-terrestrial RATs affect the latency requirements. To balance the network load in multi-RAT HetNets, we propose a distributed game-theoretic algorithm considering QoS for users. To that end, a cost function is defined to reflect the cell load status, required data rates, and delay constraints. In the proposed algorithm, overloaded cells iteratively minimize user cost in a distributed manner by associating a user with a less-loaded cell based on its cost. We provide analysis to show that minimizing each user cost in the game theory setting achieves a balanced load distribution among cells. We show that the proposed algorithm brings each cell load to a balanced state in a finite number of iterations. Via simulation, we show that the proposed algorithm achieves superior performance in terms of even load distribution, network throughput, and the number of users with satisfactory QoS.

Distributed load-balancing for account-based sharded blockchains

PurposeSharding of blockchains consists of partitioning a blockchain network into several sub-networks called “shards,” each shard processing and storing disjoint sets of transactions in parallel. Sharding has recently been applied to public blockchains to improve scalability through parallelism. The throughput of sharded blockchain is optimized when the workload among the shards is approximately the same. The purpose of this paper is to investigate the problem of balancing workload of account-based blockchains such as Ethereum.Design/methodology/approachTwo known consensus-based distributed load-balancing algorithms have been adapted to sharded blockchains. These algorithms migrate accounts across shards to balance transaction processing times. Two methods to predict transaction processing times are proposed.FindingsThe authors identify some challenging aspects for solving the load-balancing problem in sharded blockchains. Experiments conducted with Ethereum transactions show that the two load-balancing algorithms are challenged by accounts often created to process a single transaction to optimize anonymity, while existing accounts sparsely generate transactions.Originality/valueTests in this work have been conducted on transactions originating from a blockchain platform rather than using artificially generated data distributions. They show the specificity of the load-balancing problem for sharded blockchains, which were hidden in artificial data sets.

Novel distributed load balancing algorithms in cloud storage

From last decade, there is a rapid expanded upon in the data in cyberspace. In order to manage them efficiently, distributed storage came into the world. Cloud storage is a type of distributed storage based on cloud computing technology. Cloud storage acts as a repository in which data stored, managed and made accessible to users. Largest generated application datasets can flexibly be stored or deleted in the cloud and from here end users access this data by using cloud storage services interface, without accessing any storage server in real. Cloud storage system is considered of hundreds of independent storage servers which are distributed geographically, and handle millions client requests concurrently. Some of the storage servers get huge clients requests and some servers remain under loaded. Due to this unequal distribution of load on storage servers degrades the performance of overall system and increases the response time. This work addresses issues regarding to efficient utilization of storage servers in cloud storage. Handling various challenges regarding to the load balancing in the cloud storage is the one of the main objectives of this research. Though analyzing the contribution of other authors in this area, in this work two distributed load balancing algorithm CDLB and DDLB are proposed by exploiting the different parameter of storage server. The first proposed algorithm considers the service rate and queue length as a main parameter of the server. The second proposed algorithm considers extra server parameter such as service time and deadline time of the client request. This work monitors various aspects which leverage the overall performance of cloud storage. Both the algorithms try to balance the load of storage servers as well as effectively utilize the server capabilities. From the simulation results, it can be concluded that proposed algorithms balance the load, efficiently utilize the server capabilities, reduce the response time and leverage the overall system performance.

Belief Propagation for Energy Efficiency Maximization in Wireless Heterogeneous Networks

In this article, we study an energy efficient management of two-tier heterogeneous cellular networks (HetNets) which consist of one macro base station (BS) and multiple micro base stations. This article presents a distributed user association algorithm that maximizes the network-wide energy efficiency (EE) in HetNets. A subset of BSs that support only a small number of users can be turned off to save the energy consumption. By turning off BSs in the HetNet and offloading serving users to adjacent active BSs, the network-wide energy consumption is minimized, while the sum throughput is maximized. To solve the problem efficiently, we introduce a new approach based on a message-passing framework and derive a distributed load balancing algorithm. The proposed method provides a very efficient solution with reduced computational complexity compared to existing schemes. Simulation results verify that the proposed algorithm outperforms other conventional load balancing strategies.

Fairness-Aware Mechanism for Load Balancing in Distributed Systems

When a set of self-interested users shares multiple resources in a distributed system, we face the problem of allocating resources, called the load balancing problem. In particular, load balancing is defined as allocating the load to the servers of the distributed system such that jobs’ response time is minimized, and the utilization of servers is improved. In this article, the load balancing problem in a distributed system consists of a finite set of servers, and a finite set of users is studied. The load balancing problem considered here is a bi-objective problem with two highly probable conflicting objectives: (i) minimizing jobs’ response time (ii) providing the fair utilization of servers. In order to satisfy these two objectives simultaneously, both the objectives are considered in an integrated manner. Next, the load balancing problem is formulated as a noncooperative game; and to solve the game (i.e., to find the Nash equilibrium), a distributed load balancing algorithm (DLBA) is proposed. An experimental study is carried out to ascertain the efficacy of the proposed DLBA. Further, we compare DBLA with three existing load balancing approaches to evaluate its comparative effectiveness. The experimental results validate the effectiveness of the DLBA over the existing approaches.

Distributed load balancing frequent colossal closed itemset mining algorithm for high dimensional dataset

The focus of extracting colossal closed itemsets from high dimensional biological datasets has been great in recent times. A massive set of short and average sized mined itemsets do not confine complete and valuable information for decision making. But, the traditional itemset mining algorithms expend a gigantic measure of time in mining a massive set of short and average sized itemsets. The greater interest of research in the field of bioinformatics and the abundant data across the variety of domains paved the way for the generation of the high dimensional dataset. These datasets are depicted by an extensive number of features and a smaller number of rows. Colossal closed itemsets are very significant for numerous applications including the field of bioinformatics and are influential during the decision making. Extracting a huge amount of information and knowledge from the high dimensional dataset is a nontrivial task. The existing colossal closed itemsets mining algorithms for the high dimensional dataset are sequential and computationally expensive. Distributed and parallel computing is a good strategy to overcome the inefficiency of the existing sequential algorithm. Balanced Distributed Parallel Frequent Colossal Closed Itemset Mining (BDPFCCIM) algorithm is designed for high dimensional datasets. An efficient closeness checking method to check the closeness of the rowset and an efficient pruning strategy to snip the row enumeration mining search space is enclosed with the proposed BDPFCCIM algorithm. The proposed BDPFCCIM algorithm is the first distributed load balancing algorithm to mine frequent colossal closed itemsets from high dimensional biological datasets. The experimental results demonstrate the efficient performance of the proposed BDPFCCIM algorithm in comparison with the state-of-the-art algorithms.

A distributed load balancing algorithm for deduplicated storage

While deduplication brings the advantage of significant space savings in storage, it nevertheless incurs the overhead of maintaining huge metadata. Updating such huge metadata during the data migration that arises due to load balancing activity results in significant overhead. In order to reduce this metadata update overhead, this paper proposes a suitable alternate index that tracks the data blocks even when they migrate across the nodes without explicitly storing the location information. In addition, a virtual server-based load balancing (VSLB) algorithm has been proposed in order to reduce the migration overhead. The experimental results indicate that the proposed index reduces the metadata update overhead by 74 % when compared to the existing index. Furthermore, VSLB reduces the migration overhead by 33 % when compared to the existing ID reassignment-based load balancing approach.

An Energy Efficient Clustering based Distributed Load Balancing Technique for Mobile Adhoc Networks

In a mobile adhoc network (MANET), energy efficiency and mobility prediction are the two main challenging design issues due to the mobile nature of the nodes in any direction with limited battery lifetime, thus leads to adequate topology modifications. These two issues are mainly considered to maximize the lifetime of MANET. Load-balancing and reliable data transmission among the mobile nodes is mandatory to increase the network lifetime. To achieve this, clustering techniques can be employed to minimize the topology size and to aggregate the details related to the topology. In this paper, we introduce a new clustering based distributed load balancing (D-CALB) algorithm to maximize energy efficiency and network lifetime. Furthermore, a fault tolerant feature is included in the D-CALB algorithm, which maintains a secondary CH as a backup node in case of the failure of the present CH. The presented ZXCD- CALB algorithm has undergone an extensive set of experimentation under a varying number of nodes and speed. The detailed investigation of the experimental results verified the superior nature of the presented D-CALB algorithm over compared ones under several measures.

Addressing Scalability and Storage issues in Block Chain using Sharding

There is no skepticism about the perpetuity of recorded transactions in BlockChain and its Distributed Ledger Technology. But the hype is casting shadow over the real world implementation issues. Issues like limited scalability and storage constraints impede mass adoption of this world-changing technology. In this paper, we will deliberate on the issues like scalability and storage in BlockChain Technology. We will provide solutions to overcome these problems using Sharding along with distributed load balancing algorithms. The requirement of nodes increases as the number of transactions increases in a network. Preserving the essence of BlockChain Technology, if we can cleave whole system memory into local and global Shards along with dedicated miners at local shards, we can cater to needs of scalability and storage. Distributed load balancing algorithms will be implemented in network to divide the transactions recorded in distributed public ledger among different local shards and global shards. Equal distribution of transactions at local shards will increase the resource utilization of the network. Dedicated miners on these local shards will validate the transactions, hence decreasing the network latency and overhead. Distributed Proof of Stake (PoS) Consensus protocols will validate the transactions by aggregating the results of local and global shards and will update the copy of the public ledger. To meet scalability demands, the system will expand to global shards with global miners assigned dynamically. And partitioning the ledger into local and global shards to be copied to different nodes in the network will address the storage challenges in BlockChain technology. The main contributions of this paper are: 1) Using distributed load balancing algorithms to copy local shards of distributed public ledger to different nodes dynamically with replication 2) On demand expansion of nodes from local to global shards to cater scalability 3) Distributed PoS consensus protocol executed on aggregated results to update the public ledger 4) Performance of the system will improve as load distribution of shards across the nodes and aggregation of results will work in parallel.

Non-cooperative power and latency aware load balancing in distributed data centers

In this paper we propose an algorithm for load balancing in distributed data centers based on game theory. We model the load balancing problem as a non-cooperative game among the front-end proxy servers. We model the operating cost associated with a data center as a weighted linear combination of the energy cost and the latency cost. We propose a non-cooperative load balancing game with the objective of minimizing the operating cost and obtain the structure of Nash equilibrium. Based on this structure, a distributed load balancing algorithm is designed. We compare the performance of the proposed algorithm with the existing approaches. Numerical results demonstrate that the solution achieved by the proposed algorithm approximates the global optimal solution in terms of the cost and it also ensures fairness among the users.

Flow distribution-aware load balancing for the datacenter

Leveraging centralized and distributed load balancing algorithm to jointly schedule long and short flows attracts much attention since by classifying flows, they could reduce the completion times of flows without losing the scalability. However, existing centralized-distributed joint algorithms generally classify flows according to a static threshold and are unable to adapt to traffic dynamics. More importantly, through over investigation, the overheads of the controller caused by classifying flows and handling flows can be further reduced.In this paper, we present a Flow Distribution-Aware Load Balancing (FDALB) mechanism to reduce flow completion times and achieve high scalability. In FDALB, flows are split into short flows and long flows according to a threshold. The traffic of short flows and long flows are balanced by distributed and centralized algorithms respectively. To adapt to traffic dynamics, we propose a simple yet effective scheme adaptively adjusting the splitting threshold. To reduce the overheads of classifying flows, FDALB leverages end-hosts to tag long flows, which requires no changes in networking hardware. To further reduce the overheads of handling flows, we proposed a new centralized algorithm without requiring flow rates information. Using realistic datacenter workloads, we show that FDALB reduces the average FCT of flows by up to 47% over ECMP while achieves higher scalability than the state-of-art load balancing mechanism Mahout.

Fog-Assisted Operational Cost Reduction for Cloud Data Centers

In this paper, we intend to reduce the operational cost of cloud data centers with the help of fog devices, which can avoid the revenue loss due to wide-area network propagation delay and save network bandwidth cost by serving nearby cloud users. Since fog devices may not be owned by a cloud service provider, they should be compensated for serving the requests of cloud users. When taking economical compensation into consideration, the optimal number of requests processed locally by each fog device should be decided. As a result, existing load balancing schemes developed for cloud data centers can not be applied directly and it is very necessary to redesign a cost-ware load balancing algorithm for the fog-cloud system. To achieve the above aim, we first formulate a fog-assisted operational cost minimization problem for the cloud service provider. Then, we design a parallel and distributed load balancing algorithm with low computational complexity based on Proximal Jacobian Alternating Direction Method of Multipliers (PJ-ADMM). Finally, extensive simulation results show the effectiveness of the proposed algorithm.

A distributed load balancing algorithm for climate big data processing over a multi‐core CPU cluster

SummaryLoad imbalance is a common problem to be tackled urgently in large scale data‐driven simulation systems or data intensive computing. According to the coupler, the Chinese Academy of Sciences‐Earth System Model (CAS‐ESM) implements one‐way nesting of the Institute of Atmospheric Physics of Chinese Academy of Sciences Atmospheric General Circulation Model version 4.0 (IAP AGCM4.0) and Weather Research and Forecasting model (WRF). The METGRID (meteorological grid) and REAL program modules in the WRF are used to process meteorological data. In the CAS‐ESM, the load of the METGRID module is seriously unbalanced on many CPU cores. The load imbalance has a serious impact on the processing speed of meteorological data, so this study designs an optimization algorithm to solve the problem. Numerical experiments show that compared to before optimization, the optimization algorithm can solve the load imbalance of the METGRID, and the computation speed of the METGRID and REAL modules after optimization on 64 CPU cores is about 7.2 times faster than before. Meanwhile, the whole computation speed of the CAS‐ESM can improve by 217.53%. In addition, results indicate that they also can reach to a similar speedup on different numbers of CPU cores. Copyright © 2016 John Wiley & Sons, Ltd.

Comparative Analysis of Existing Dynamic Load Balancing Techniques

The anticipated uptake of Cloud computing, built on well- established research in Web Services, networks, utility computing, distributed computing and virtualization, will bring many advantages in cost, flexibility and availability for service users. Cloud is based on the data centers which are powerful to handle large number of users. As the cloud computing is a new style of computing over internet, it has many advantages along with some crucial issues to be resolved in order to improve reliability of cloud environment. Central to this is the implementation of an effective load balancing algorithm. This paper investigates two distributed load balancing algorithms which have been proposed for load balancing: round robin and throttled scheduling.

Periodic hierarchical load balancing for large supercomputers

Large parallel machines with hundreds of thousands of processors are becoming more prevalent. Ensuring good load balance is critical for scaling certain classes of parallel applications on even thousands of processors. Centralized load balancing algorithms suffer from scalability problems, especially on machines with a relatively small amount of memory. Fully distributed load balancing algorithms, on the other hand, tend to take longer to arrive at good solutions. In this paper, we present an automatic dynamic hierarchical load balancing method that overcomes the scalability challenges of centralized schemes and longer running times of traditional distributed schemes. Our solution overcomes these issues by creating multiple levels of load balancing domains which form a tree. This hierarchical method is demonstrated within a measurement-based load balancing framework in Charm++. We discuss techniques to deal with scalability challenges of load balancing at very large scale. We present performance data of the hierarchical load balancing method on up to 16,384 cores of Ranger (at the Texas Advanced Computing Center) and 65,536 cores of Intrepid (the Blue Gene/P at Argonne National Laboratory) for a synthetic benchmark. We also demonstrate the successful deployment of the method in a scientific application, NAMD, with results on Intrepid.

The Convergence of Realistic Distributed Load-Balancing Algorithms

We give a general model of partially asynchronous, distributed load-balancing algorithms for the discrete load model in parallel computers, where the processor loads are treated as non-negative integers. We prove that all load-balancing algorithms in this model are finite. This means that all load-balancing algorithms based on this model are guaranteed to reach a stable situation at a certain time (which depends on the particular algorithm) at which no load will be sent from one processor to another. With an additional assumption, we prove that the largest load difference between any two processors, in the final stable situation of the load-balancing algorithms in this model, is upper-bounded by the diameter of the topology.

Design and Performance Evaluation of Queue-and-Rate-Adjustment Dynamic Load Balancing Policies for Distributed Networks

In this paper, we classify the dynamic distributed load balancing algorithms for heterogenous distributed computer systems into three policies: queue adjustment policy (QAP), rate adjustment policy (RAP), and queue and rate adjustment policy (QRAP). We propose two efficient algorithms, referred to as rate-based load balancing via virtual routing (RLBVR) and queue-based load balancing via virtual routing (QLBVR), which belong to the above RAP and QRAP policies, respectively. We also consider algorithms estimated load information scheduling algorithm (ELISA) and perfect information algorithm, which were introduced in the literature, to implement QAP policy. Our focus is to analyze and understand the behaviors of these algorithms in terms of their load balancing abilities under varying load conditions (light, moderate, or high) and the minimization of the mean response time of jobs. We compare the above classes of algorithms by a number of rigorous simulation experiments to elicit their behaviors under some influencing parameters, such as load on the system and status exchange intervals. We also extend our experimental verification to large scale cluster systems such as a mesh architecture, which is widely used in real-life situations. From these experiments, recommendations are drawn to prescribe the suitability of the algorithms under various situations

Dual and multiple token based approaches for load balancing

In distributed systems uneven arrivals of the tasks may overload a few hosts while some of the hosts may be lightly loaded. This load imbalance prevents distributed systems from delivering its performance to its capacity. Load balancing has been advocated as a means of improving performance and reliability of distributed systems. We propose a distributed load balancing algorithm LoGTra to deal with this problem. LoGTra uses load graph and token based policy. The extensions to LoGTra based on dual tokens DTLB and multiple tokens m - LoGTra are proposed in this paper. m - LoGTra allows host to generate multiple tokens. This allows system to search host for load balancing in multiple directions and can avoid starvation of remote hosts. Local maxima and local minima are responsible for initiating transfer and that limits the number of hosts generating tokens at a time. As overheads are kept under control by limiting number of tokens, the algorithm promises for improved performance.