Balancing Server Research Articles

CLIP-Guided Federated Learning on Heterogeneity and Long-Tailed Data

Federated learning (FL) provides a decentralized machine learning paradigm where a server collaborates with a group of clients to learn a global model without accessing the clients' data. User heterogeneity is a significant challenge for FL, which together with the class-distribution imbalance further enhances the difficulty of FL. Great progress has been made in large vision-language models, such as Contrastive Language-Image Pre-training (CLIP), which paves a new way for image classification and object recognition. Inspired by the success of CLIP on few-shot and zero-shot learning, we use CLIP to optimize the federated learning between server and client models under its vision-language supervision. It is promising to mitigate the user heterogeneity and class-distribution balance due to the powerful cross-modality representation and rich open-vocabulary prior knowledge. In this paper, we propose the CLIP-guided FL (CLIP2FL) method on heterogeneous and long-tailed data. In CLIP2FL, the knowledge of the off-the-shelf CLIP model is transferred to the client-server models, and a bridge is built between the client and server. Specifically, for client-side learning, knowledge distillation is conducted between client models and CLIP to improve the ability of client-side feature representation. For server-side learning, in order to mitigate the heterogeneity and class-distribution imbalance, we generate federated features to retrain the server model. A prototype contrastive learning with the supervision of the text encoder of CLIP is introduced to generate federated features depending on the client-side gradients, and they are used to retrain a balanced server classifier. Extensive experimental results on several benchmarks demonstrate that CLIP2FL achieves impressive performance and effectively deals with data heterogeneity and long-tail distribution. The code is available at https://github.com/shijiangming1/CLIP2FL.

Power Metering Automation System Based on Internet of Things

In order to solve the problems of poor stability and low operation efficiency of the traditional power marketing metering production automatic scheduling system, this paper proposes a power metering automation system based on the Internet of Things. In this paper, the hardware and software of the system are designed under the environment of the Internet of Things. The hardware part is mainly divided into three platforms: basic resource layer, platform business layer, and interface presentation layer. The normal operation is realized through the network topology, and the load balancing server and business subsystem are emphatically designed. The software is completed in four steps: coding, scheduling, backup, and display. The experimental results show that the integrity rate of the electric quantity data collected by this system in the experiment with the traditional system is as high as 99%. Compared with the traditional dispatching system, the power information dispatching system designed under the Internet of Things has better stability and higher operation efficiency.

Discrete-event simulation of load distribution between server stations

The simulation model of the computational load balancing in a server complex using a balancing server is proposed. The server load balancing model was developed using MATLAB/Simulink/SimEvents/Stateflow tools. The SimEvents based model makes it possible to simulate the server complex as a queuing system and to take into account the sporadic occurrence of requests. The Stateflow based event-driven model simulates the system with a variable time delay in data transmission channels and simulates state transitions for servers. This discrete-event simulation model allows evaluating the possibility of rational use of computational resources and reducing the time of service. In addition, the model allows to investigate the effectiveness of load balancing systems at the stage of their development.

Effectively Handling Network Congestion and Load Balancing in Software-Defined Networking

The concept of Software-Defined Networking (SDN) evolves to overcome the drawbacks of the traditional networks with Internet Protocol (I.P.) packets sending and packets handling. The SDN structure is one of the critical advantages of efficiently separating the data plane from the control plane to manage the network configurations and network management. Whenever there are multiple sending devices inside the SDN network, the OpenFlow switches are programmed to handle the limited number of requests for their interface. When the recommendations are exceeded from the specific threshold, the load on the switches also increases. This research article introduces a new approach named LBoBS to handle load balancing by adding the load balancing server to the SDN network. Besides, it is used to maximize SDN’s reliability and efficiency. It also works in coordination with the controller to effectively handle the load balancing policies. The load balancing server is implemented to manage the switches load effectively. Results are evaluated on the NS-3 simulator for packet delivery, bandwidth utilization, latency control, and packet decision ratios on the OpenFlow switches. It has been found that the proposed method improved SDN’s load balancing by 70% compared to the previous state-of-the-art methods.

Performance Analysis of a Cloud Computing System using Queuing Model with Correlated Task Reneging

Queuing theory has been extensively used in the modelling and performance analysis of cloud computing systems. The phenomenon of the task (or request) reneging, that is, the dropping of requests from the request queue often occur in cloud computing systems, and it is important to consider it when developing performance evaluations models for cloud computing infrastructures. Majority of studies in the performance evaluation of cloud computing data centres with the use of queuing theory do not consider the fact that the tasks could be removed from queue without being serviced. The removal of tasks from the queue could be due to the user impatience, execution deadline expiration, security reasons, or as an active queue management strategy. The reneging could be correlated in nature, that is, if a request is dropped (or reneged) at any time epoch, and then there is a probability that a request may or may not be dropped at the next time epoch. This kind of dropping (or reneging) of requests is referred to as correlated request reneging. In this paper we have modelled a cloud computing infrastructure with correlated request reneging using queuing theory. An M/M/1/N queuing model with correlated reneging has been used to study the performance analysis of the load balancing server of a cloud computing system. The steady-state as well as the transient performance analyses have been carried out. Important measures of performance like average queue size, average delay, probability of task blocking, and the probability of no waiting in the queue are studied. Finally, some comparisons are performed which describe the effect of correlated task reneging over simple exponential reneging.

Load Balancing Server and Homomorphic Encryption in Internet of Things

User demand for Internet of Things (IoT) services is ever increasing. The growing user demand can lead to an escalation of server workloads and faces the threat of theft of critical data. Consequently, a system is necessary to balance the server load and is protected with data encryption. In this study, we designed a system to share server workloads using load balancing methods. The load balancing technique uses open-source web server software. The system is equipped with data security using a homomorphic encryption algorithm from AES on the sender's side. The system embeds in an IoT telemedicine apparatus. During testing, we analyze the error requests that arrive at each server for the HTTP GET and POST methods. We also evaluate the speed of data encryption and decryption. The results showed that server load balancing reduces the number of error requests for the GET method by 97%. Meanwhile, the number of error requests for the POST method decreases by 66.75%. Observations reveal that the average homomorphic encryption speed, computation time, and decryption time are 15.66 ms, 764.18 µs, and 362.49 µs, respectively.

Software-Defined Networking (Sdn) In Modern Network Management

Modern computer network implements tasks from routing, monitoring traffic, accessing control and balancing server load. Software-Defined Networking (SDN) has become a new radical model where the network is programmable. The network can be simplified and managed through network programmability. In addition, SDN offers a very good interface between networking devices and the software that control this activity. This paper presents the SDN architecture, ForCES, and OpenFlow protocol to allocate information exchange. Oracle and HP is presented as an organization that built SDN product by having different architecture. Always as a new solution for computer network has been proposed, it still has challenges for researcher to solve the SDN issues that has been raised. Keywords: Software-Defined Networking (SDN) , programmable networks , ForCES , OpenFlow

Software-Defined Networking (Sdn) In Modern Network Management

Modern computer network implements tasks from routing, monitoring traffic, accessing control and balancing server load. Software-Defined Networking (SDN) has become a new radical model where the network is programmable. The network can be simplified and managed through network programmability. In addition, SDN offers a very good interface between networking devices and the software that control this activity. This paper presents the SDN architecture, ForCES, and OpenFlow protocol to allocate information exchange. Oracle and HP is presented as an organization that built SDN product by having different architecture. Always as a new solution for computer network has been proposed, it still has challenges for researcher to solve the SDN issues that has been raised. Keywords: Software-Defined Networking (SDN), programmable networks, ForCES, OpenFlow

Load balancing in data distribution systems

The issue of distribution of large amounts of information on server stations is researched. The analysis of existing methods for distributing data of heavily loaded web applications is carried out. A system for computing load distribution using a balancing server is proposed. A description of the software necessary for the functioning of the system and server management interface is provided.

РАСПРЕДЕЛЕНИЕ ДАННЫХ В ИНФОРМАЦИОННОЙ СИСТЕМЕ С ПОМОЩЬЮ СЕРВЕРА-БАЛАНСИРА

The issue of distribution of large amounts of information system information on server stations is considered. The analysis of existing methods for distributing data of highly loaded web applications is carried out. A system for computing load distribution using a balancing server is proposed. A description of the software necessary for the functioning of the system, and a server management interface are provided.

Adaptive Load Balancing Strategy Based on LVS

Linux Virtual Server (LVS) is a load balancing server that deployed on a cluster of real servers and the load balancer running on the Linux operating system. In the scheduling module, we designed an adaptive load balancing strategy. Firstly, we introduce the traffic state prediction. The algorithm we choose is Radial Basis Function (RBF) neural network. Secondly, we update weights according to the real-time usage of CPU, memory and hard disk. Then we conduct a series of experiments and the adaptive strategies have a better performance on load balancing.

로드 밸런싱 Active-Active 방식의 웹 서버 이중화 구축 및 결함내성 시험

본 논문에서는 Active-Active 방식의 웹서버 이중화 방법에 관해 연구하였다. Rsync과 crontab을 이용하여 웹서버간에 주기적인 복사가 이루어지도록 하여 동일한 상태를 유지시켰다. 로드밸런싱서버가 웹서버를 교대로 실행하여 부하를 분산하고, 빠른 서비스가 제공되도록 하였다. 또한 임의의 웹서버에 장애가 발생하더라도 또 다른 웹서버에 의해 서비스가 계속 유지되도록 하였다. In this paper we researched on the Duplication Techniques for Active-Active Web Servers. Rsync and crontab utilities make copy periodically between web servers and maintain the same status. Load Balancing Server makes web servers load balanced and fast servicing by executing web servers alternatively. Even though one web server stops due to some critical errors, the remaining web server can take over and provide services continuously.

Quantitative survivability evaluation of three virtual machine-based server architectures

Virtual machine-based services have become very popular in data centers and cloud computing in recent years. Efficient redundancy technology of virtual machine provides good availability for services; thus, it has become possible to provide continuous services even if the system suffers intrusions. So far, many intrusion tolerant server architectures have been developed based on virtual machine technology in academia and industry. Unfortunately, to the best of our knowledge, there is very little work done on evaluating the survivability of virtual machine-based server architectures in the literature.In this paper, we focus on analyzing and evaluating the survivability of three virtual machine-based architectures, which are load balance server architecture (LBSA), isolated component server architecture (ICSA), and Byzantine fault tolerant server architecture (BFTSA). As a reference, a traditional server cluster without virtual machines is also analyzed and compared. We model different architectures with Continuous Time Markov Chain (CTMC). We analyze the transient behaviors and steady states of different virtual machine-based architectures. Furthermore, the costs of the traditional server cluster and the virtual machine-based architectures are compared in terms of processing costs, memory costs, communication costs, and fail-safe fault tolerance. The results show that BFTSA has better survivability than LBSA and ICSA, but with longer time to reach the steady states and higher communication costs.

Study on virtual machine’s live migration

Virtual machine technology is a good way to separate software and hardware in the same physical machine,and is beneficial to fault management,dynamic load balancing and online server maintenance.It is especially useful in optimizing the performance of data center and cluster.To compare the performance of stop and copy migration and live migration,as well as to compare the impact of different tasks on live migration,this paper migrated four kinds of virtual machines with different tasks in LAN,using XEN as a virtual machine monitor.

Optimizing server placement in hierarchical grid environments

In this paper, we address some problems related to server placement in Grid environments. Given a hierarchical network with requests from clients and constraints on server capability, the minimum server placement problem attempts to place the minimum number of servers that satisfy requests from clients. Instead of using a heuristic approach, we propose an optimal algorithm based on dynamic programming to solve the problem. We also consider the balanced server placement problem, which tries to place a given number of servers appropriately so that their workloads are as balanced as possible. We prove that an optimal server placement can be achieved by combining the above algorithm with a binary search on workloads. This approach can be further extended to deal with constrains on network capability. The simulation results clearly show the improvement in the number of servers and the maximum workload. Furthermore, as the maximum workload is reduced, the waiting time is reduced accordingly.

A LOW LATENCY PROXY PREFETCHING CACHING ALGORITHM

The Web proxy cache system was deployed to save network bandwidth, balance server load, and reduce network latency by storing copies of popular documents in the client and proxy caches for the Uniform Resource Locator (URL) requests. To solve the problem of the Web's slow end-user response time, a Web proxy caching and prefetching strategy has been developed and implemented by the auother to provide the users by the information they mostly likely want to browse in user profiles. This developed strategy uses the Reverse Aggressive technique for prefetching, which was proposed theortically. This developed strategy has been implemented with different cache sizes using a Web caching simulator. The tradional caching replacement policies such as Least-Recently-Used (LRU), Hybrid, and Size policies were already existed in this simultor. This simulator has been modified by the work in this paper such that the most recent replacement policies; Last-In-First-Out (LIFO), First-Try, Swapping and Place-Holder policies under infinite sized cache have been implemented. The performance measurements of the developed strategy have been studied using the tradional replacement policies, and the most recent replacement policies. Also, a comparative study has been done to clarify the benefits of the Reverse Aggressive caching prefetching algorithm with respect to the Fixed-Horizon caching prefetching algorithm with respect to the Reduced Latency (RL). According to the implementation results, it has been found that the average latency has been reduced at a higher degree by using the Reverse Aggressive cache prefetching strategy.