Hemant Kumar Mehta

On trust management and reliability of distributed scheduling algorithms

This paper presents a mechanism of integrating the concept of reliability with trust management policy in the distributed scheduling algorithms. This algorithm focuses on the distributed computing systems involving collaborative computing. Trust management is achieved with the concept of bidirectional reputation points that are assigned to resource providers and resource consumers. Moreover, concept of reliability is also used to ensure failsafe execution of the processes. This is a generalized mechanism that can be integrated with existing distributed scheduling algorithms used in collaborative computing environment viz. Grid computing, cloud computing, peer-to-peer computing etc. Extensive simulations were performed to test the applicability of the mechanism. Our results confirm that the proposed method significantly reduces the number of failures as compared to a distributed scheduling algorithm implemented without reliability and trust management policy.

Estimating the Power Consumption of an Idle Virtual Machine

Power management has become one of the main challenges for data center infrastructures. Currently, the cost of powering a server is approaching the cost of the server hardware itself, and, in a near future, the former will continue to increase, while the latter will go down. In this context, virtualization is used to decrease the number of servers, and increase the efficiency of the remaining ones. If virtualization can be used to positively impact on the data center energy consumption, this new abstraction layer disconnects user services (hosted on a virtual machine) from their operating cost. In this paper, we propose an approach and a model to estimate the total power consumption of a virtual machine, by taking into account its static (e.g. memory) and dynamic (e.g. CPU) consumption of resources. This model permits to reconnect each VM to its corresponding operating cost, and provides more information to virtual infrastructure providers and users to optimize their infrastructure/applications. It can be observed from results of experiments that the proposed method outperforms the methods found in the literature that only consider the dynamic consumption of resources.

Performance evaluation of Grid simulators using profilers

Profilers are the performance analysis tools that produce the traces of the execution of the applications developed in various programming languages. These traces can be analyzed to determine the performance of the applications. This paper compares the performance of two Grid simulators developed in Java namely the EcoGrid (developed by the authors of the paper) and GridSim. The profiler tool available with the NetBeans integrated development environment has been used. The scalability, memory usage and the runtime performance of the simulators are tested on the profilers. We showed that EcoGrid has much better performance as compared to GridSim.

EcoGrid: a dynamically configurable object oriented simulation environment for economy-based grid scheduling algorithms

In this paper, an economy-based Grid simulator named as EcoGrid is presented. EcoGrid is developed as a test-bed for Grid scheduling algorithms that are based on dynamic load balancing. EcoGrid is dynamically configurable. It optimizes the cost of execution of a process and maximizes the profit of a service provider system. As compared to other commonly used Grid simulation tools like GridSim and SimGrid, this simulator provides more enhanced features like configurable, scalable and extensible. This object oriented simulator also supports execution of Java applications. EcoGrid supports simulation of both the economy based and non-economy based scheduling algorithm.

Towards development of a distributed e-Learning ecosystem

This paper explores the possibility of application of cluster, grid and cloud computing in developing e-Learning systems. We are considering three major problems in implementation of distributed e-Learning systems: (i) a method for selection of appropriate distributed computing system (DCS) e.g. cluster, grid and cloud. (ii) To find suitable resource management policies. (iii) To propose an architecture for distributed e-Learning system.

A Modified Delay Strategy for Dynamic Load Balancing in Cluster and Grid Environment

This paper considers a specific Cluster and Grid computing environment, where the workstations are under full load situation. In such situations, we delay the scheduling of new jobs instead of dispatching them to one of the overloaded workstations. We present a modified delay strategy mDELAY, to significantly enhance delay-based scheduling algorithm proposed by Hui & Chanson and named as DELAY. The proposed mDELAY algorithm produces better batch completion time and improved load balancing at workstations. The experiments are performed on a centralized scheduler. However to improve the performance a decentralized scheduling framework based on web service or Grid service is also proposed.

Economy Based Resource Allocation in IaaS Cloud

Infrastructure as a Service IaaS offers hardware resources computing power, storage and network as a service to its customers. The customers order these resources in the form of a lease. Aim of any service provider is to make a leasing plan to maximize the number of accepted leases. Opennebula is popular open source toolkit for building IaaS cloud. Opennebula has its own lease manager and it can also be integrated with Haizea which is an open source lease manager. An economy based algorithm should focus on incentives of both the consumers and the providers. In this paper, an economy based leasing algorithm is developed and integrated with Haizea. This economy based algorithm takes care of incentives of both the parties i.e. customer and service provider. It uses the concept of optimization techniques to optimize the costs. The incentive for customers is lower cost of execution of its lease on the capable node as compare to existing non-economy based algorithms. If an appropriate resource is not found, then the algorithm uses negotiation on budget and resource demand; that increases the number of accepted lease. Thus, incentive for providers is an increase in profit as the amount of accepted leases increase. Experimental results show that the proposed economy based leasing algorithm reduces the cost of execution of the consumers lease and increases the profit of the provider to a considerable extent.

Maximum Utility Meta-Scheduling Algorithm for Economy Based Scheduling under Grid Computing

In this paper, an economy-based Grid algorithm named as maximum utility (MU) has been presented. The algorithm maximizes the utility for all of the users. The maximum utility for the resource provider indicates the higher profit and for the resource consumers the maximum utility indicates successful execution of users processes with the lower cost of execution. Thus the utility results in the better incentives to the users. The algorithm uses the market based parameters to determine the utility value for both the resource consumers and the resource providers. Two parameters, loss factor and net savings, have been used to characterize the utility values for resource providers and resource consumers. Besides cost of execution, the algorithm uses the past performance of resource providers before allocating a new process. The algorithm uses a parameter called reputation point that will be credited to the resource provider’s account after successful completion of a process. A hierarchical decentralized scheduling framework is also proposed for better performance of scheduling algorithms.

Distributed database caching for web applications and web services

The demand and usage of web application are increasing exponentially. Generally, these web applications are dynamic in nature. These applications retrieve the data from databases based on users requests. The high demand servers receive thousands of requests in a second. This overloads the database and degrades the overall performance of the web applications. If the database and web server are implemented on the same server, the database may become the bottleneck. Hence, these applications are implemented as multitier application. In multitier web applications and web services the web based frontend, business logic and databases are deployed over the different servers. Although this will introduce the network delay as web server, application server and database server are connected through network. To avoid the network delay and to improve the performance of multitier applications a new distributed hash-based database cache implementation is presented. Centralized control of the distributed cache is implemented for overall improved performance.

A Two Level Broker System for Infrastructure as a Service Cloud

The potential market of cloud computing has encouraged various providers to offer their services on the cloud platform. This ever growing number of cloud providers is resulting in the demand of the brokering services for cloud computing providers as the consumers requires support in acquiring appropriate services for their functionality. It require a lot of time and efforts in analyzing the market and information about various service providers. To cope up with this problem it is better for consumers to use a brokering service to finalize the cloud service provider. In last few years, the researchers have focused on developing brokering services for cloud computing environment. Mostly these brokers performs the matching of consumers requirement and facilities provided by provider. Some of the brokers also performs the ranking of providers to help the consumers. However, these broker performs this matching only on the basis of the demand of cloud resources such as processor, RAM, disk space, bandwidth etc. We call these parameters as standard attributes. In this paper, we proposed a multipurpose cloud broker system to search cloud resources with the help of standard and custom attributes and control these resources. The cloud broker provides an optimized list of cloud resources based on selected attributes to help the consumer in selection of most suitable resources. Two level architecture is used to develop this cloud broker system. At level one, there is a master broker, which consists of three modules namely registration, searching and controlling. At the second level, there is a local broker on each resource provider site. It provides information of resources to master broker and help consumers to control their leased resources. This local broker works under control of master broker. Extensive experiments have been performed to prove the effectiveness of the proposed approach. From the results of experimentations, it is evident that proposed broker helps consumers to get best available resources. Currently, we are enhancing this broker so that it can also perform resource management operations for intercloud environment.