R. K. Pateriya

Cloud Server Optimization with Load Balancing and Green Computing Techniques Using Dynamic Compare and Balance Algorithm

Cloud computing is a business oriented concept to provide online IT resources and IT services on demand using pay per use model where main goal of cloud service provider is to use cloud computing resources efficiently and gain profits marginally. One of the challenging areas in cloud computing is frequent optimization of cloud server. It mainly concerns with the load balancing of cloud data centers to improve efficiency of the host machine and minimize number of active host machine to support green computing concept. To balance the load of entire data center, we need to transfer the virtual machines of the overloaded host to the light weighted host using migration techniques. In this paper, we introduce a threshold based Dynamic compare and balance algorithm (DCABA) for cloud server optimization. Unlike the traditional server optimization strategies which consider only load balancing and scheduling of resources based on the usage of CPU, RAM and BW in physical servers, DCABA also minimizes the number of host machines to be powered on, for reducing the cost of cloud services. Our approach can serve the purpose of service cost reduction in cloud industry with effective utilization of available resources.

Balance Resource Utilization BRU Approach for the Dynamic Load Balancing in Cloud Environment by Using AR Prediction Model

Oneofthemajorchallengesforthecloudprovideristheefficientutilizationofthephysicalresources. Toachievethis,thispaperproposedaBalanceResourceUtilization(BRU)approachthatnotonly minimizestheresourceleakagebutalsoincreasestheresourceutilizationandoptimizethesystem performance.Theproposedapproachconsider tworesources i.e.,CPUandmemory,asdecision metrics for loadbalancinganduse three thresholdsnamed lower threshold,upper thresholdand warningthresholdtodefineunderloaded,overloadedandwarningsituations,respectively.Themain conceptof thisapproach is toplaceVMto thePM,whereresourcerequirementof theVMand resourceutilizationofthePMarecomplementstoeachother.Toevadeunnecessarymigrationsdue tothetemporarypeakloadARtimeseriespredictionmodelisused.Theauthors’approachtreatsload balancingproblemfromthepracticalperspectiveandimplementedinOpenStackcloudwithKVM hypervisor.Moreover,proposedapproachresolvetheissueofVMmigrationintheheterogeneous environment. KEywORDS Auto Regression (AR) Model, CPU Load, CPU Utilization, Energy Efficient, Response Time, Virtual Machine, Warning Threshold

Web server load management with adaptive SSL and admission control mechanism

Today internet based e-commerce has become a trend and business necessity. Security of online e-commerce over the Internet is provided through Secure Sockets Layer / Transport Layer Security (SSL/TLS). The SSL/TLS uses public key cryptographic algorithms which are computational intensive due to the modular multiplications. Therefore, SSL/TLS servers often become overloaded while performing public key decryptions when the simultaneous requests increase rapidly. Overload can lead e-commerce applications to considerable revenue losses, response times may grow to unacceptable levels and as a result the server may saturate even crash. So load management of SSL/TLS web server is a critical issue in e-commerce. The proposed work in this paper is to manage load on the server by applying Adaptive SSL policy and session handling mechanism. This policy is self-adaptive security, which offers great potential in providing timely and fine grained security control on server, by runtime adaptation of cryptographic algorithm. The implementation results in more efficient system with better throughput and response time.

Shilling Attack Detection in Recommender System Using PCA and SVM

Shilling Attack is where, deceptive users insert fake profiles in recommender system to bias the rating, which is termed as shilling attack. Several studies were conducted in past decade to scrutinize different shilling attacks strategies and their countermeasures mainly categorized in linear algebra, statistical, and classification approach. This paper explores two different methods for shilling attack detection namely, Principal Component Analysis (PCA) and Support Vector Machine (SVM) and compares their performance on attack detection. We had experimented with simulating various attack models like average, random, and bandwagon models. This paper further discusses the importance of detecting malicious profile from the genuine one and suggests deep insights of developing new and more efficient shilling attack detection techniques. The experiments were conducted on the Movie Lens 100 K Dataset and compared the performance of PCA technique with supervised SVM-classification method.