• No results found

Task Assignment Policy Using Min-Max Algorithm for A Distributed Server System

N/A
N/A
Protected

Academic year: 2020

Share "Task Assignment Policy Using Min-Max Algorithm for A Distributed Server System"

Copied!
5
0
0

Loading.... (view fulltext now)

Full text

(1)

59

Abstract:

Abstract: -Data in the cloud can be shared by multiple users and is the place for large storage. Hybrid Clouds are a composition of two or more clouds (private, community or public). Hybrid cloud computing model is designed to reduce workload using max-min scheduling Algorithm. Each task executed by geographically distributed servers, so that service availability ratio is high even if the number of request is more. Each request is executed by different server depends on the response time of the server. The response time of each task is calculated by RT=RST-RRT. The response time and bandwidth information of all the servers are stored in this information server and it is used to reduce the maintenance workload. The core technology of the intelligent workload factoring service is a fast frequent data item detection algorithm, which enables factoring incoming requests not only on volume but also on data content, upon changing application data popularity. The server will get the minimum response time form the information server. If the server has minimum response time then it will be assigned to execute the first task and so on. Thus the workload is effectively managed by dividing the tasks into subtasks and then the tasks are executed based on the response time.

1. Introduction:

Cloud Computing technology portfolio behind such services, features a shared elastic computing infrastructure hosting multiple applications where IT management complexity is hidden and resource multiplexing leads to efficiency; more computing resources can be allocated on demand to an application when its

current workload incurs more resource demand than it was allocated. Despite the advantages at management simplification and pay-per-use utility model, Cloud Computing remains in doubt regarding enterprise IT adoption[1]. The concerns on the current cloud computing services include service availability & reliability, lack of Service Level Agreements, customer data security & privacy, government compliance regulation requirements.

The deployment of individual workloads and workload-based business services in the "hybrid distributed data center including physical machines, data centers, private clouds, and the public cloud raises a host of issues for the efficient management of provisioning, security, and compliance[2]. By making workloads "intelligent" so that they can effectively manage themselves in terms of where they run, how they run, and who can access them, intelligent workload management addresses these issues in a way that is efficient, flexible, and scalable.

A workload is considered "intelligent" when it a) understands its security protocols and processing requirements so it can self-determine whether it can deploy in the public cloud, the private cloud or only on physical machines; b) recognizes when it is at capacity and can find alternative computing capacity as required to optimize performance; c) carries identity and access controls as well as log management and compliance reporting capabilities with it as it moves across environments; and d) is fully integrated with the business service management layer, ensuring that end user computing requirements are not disrupted by distributed computing resources, and working with current and emergent IT management frameworks[3].

Fig.1 Dynamic workload management system

This work addresses several key challenges pertaining to dynamic workload

Task Assignment Policy Using Min-Max Algorithm for A

Distributed Server System

N.Gokila1 V.Gowdhaman2 1. M.E., Students, ngokila85@gmail.com 2. Assistant Professor, vsgowdham@gmail.com

(2)

60

management in heterogeneous Cloud environments. Specifically, we first present a scheme as shown in fig.1 that place service application across geographically distributed data centers to meet service demand while minimizing total resource usage cost. Then, we design a heterogeneity-aware dynamic application provisioning technique to minimize energy consumption while satisfying performance objectives. The problem of MapReduce scheduling and present a novel scheme that leverages heterogeneous run-time task usage characteristics. Through experiments and simulations, we show our proposed solutions can significantly reduce data center energy consumption, while achieving better application performance in terms of service response time and job completion time.

2. Related work:

Related work was scaled out by Hui Zhang et al [5] about the problem of workload management. So a task scheduling is proposed and based on the response time o the servers the workload is efficiently managed. Thus the tasks are divided into subtasks and then allocated to the server which has good response time. All the information are stored in the repository or future purpose.

2.1 Problem Statement:

Despite the advantages at management simplification and pay-per-use utility model, Cloud Computing remains in doubt regarding enterprise IT adoption. The concerns on the current cloud computing services include service availability &reliability, lack of Service Level Agreements, customer data security & privacy, government compliance regulation requirements, and more [6]. The unexpected spikes in the workload cannot be predicted so it is necessary to learn the nature of the business and find out an efficient way to handle it once such events happen.

Load balancing distributes workloads across multiple computing resources. But Load Unbalancing is reduces resource use, minimize throughput, maximize response time, and overload of any single resource. Using single components with load balancing instead of a single component may reduce reliability through redundancy. Service availability is the biggest problem in existing system while the number of request increases dramatically[7]. 2.2 Design Objectives:

This project will improve the cloud server performance by partitioning the task to sub task and then allocating the subtask to distributed sub servers. While the motivation of the hybrid cloud computing model originates from dynamic workload management, it addresses many concerns on the full Cloud Computing model where customers completely rely on public cloud services. The service availability ratio is high even if the number of request is more.

2.3 Algorithm and Methodologies: (1) Task Scheduling Algorithm

Task scheduling process is an allocation of one or more time intervals to one or more resources. In cloud computing, the scheduling is a problem of scheduling a set of submitted tasks from different users on a set of computing resources to minimize the completion time of a specific task or the makespan of a system[4][8]. There are many other parameters can be mentioned as factor of scheduling problem to be considered such as load balancing, system throughput, service reliability, service cost, system utilization and so forth. Through comprehensive study of scheduling, Task scheduling algorithm is a decision making process about assigning and finding the best match between tasks and resources.

(2) Max-min scheduling algorithm

Min-Min and Max-Min algorithms are common applicable in small scale distributed systems. When the number of small tasks is more than number of the large tasks in a meta task, the Max-min algorithm schedules tasks, in which the makespan of the system relatively depends on how many, executing small tasks concurrently with large one. If can't execute tasks concurrently, makespan become large.

(3)

61

3. Proposed Work:

An intelligent workload factoring service is designed as an enabling technology of the hybrid cloud computing model. Its basic function is to split the workload into two parts- upon (unpredictable) load spikes, and assures that the base load part remains within plan in volume, and the flash crowd load part incurs minimal cache/replication demand on the- application data required by it. This simplifies the system architecture for the flash crowd load zone and significantly increases the server performance within it[10]. As for the base load zone, workload dynamics are reduced significantly; this makes possible capacity planning with low over-provisioning factor and/or efficient dynamic provisioning with reliable workload prediction. We focus on the workload factoring component in this paper as it is a unique functionality requirement in the hybrid cloud computing architecture.

The workload factoring service for enterprise customers to make the best use of public cloud services along with their privately-owned (legacy) data centers. It enables federation between on- and off-premise infrastructures for hosting Internet-based applications, and the intelligence lies in the explicit segregation of base workload and trespassing workload, the two naturally different components composing the application workload[11]. The core technology of the intelligent workload factoring service is a fast frequent data item detection algorithm, which enables factoring incoming requests not only on volume but also on data content, upon changing application data popularity.

3.1 System Architecture:

Fig.2 System Architecture

Many more computing resources will be available at different geographical locations which is shown in fig.2. To minimize the response time of requests, application servers closer to the user and execute the task. To assign the task for each server which depends on the response time of server and the main server allocate the task to sub server. Response time of the entire available server is calculated to find the minimum response time. The response time of previous task history is maintained in the information repository. The response time of new task is taken according to the previous task. If the server has minimum response time then it will be assigned to execute the first task and so on.

3.2 Modules:

(i) User Login

The module has two categories as administrator and user play an important position. Administrator has the responsibility of creating an account to the new user and maintaining the account information of all the existing users. To create a new account the user has to give their information to the administrator. If it is valid information then the account will be created by the administrator to continue the transaction. If an account is created successfully, the existing user can view their own account information and they can transfer amount to another account.

(ii) Response Time Calculation

The user can request the data to the server. If the server has the information related to the user request, it will respond the user. The response time of each task is calculated by RT=RST-RRT. Here RT denotes response time of the requested task, RST denotes request starting time and RRT denotes response receiving time. The response time of each and every server may vary from one server to another server. Here the response time of the entire available server is calculated to find the minimum response time.

(iii) Information Repository

(4)

62

time of previous task history is maintained in the information repository. The response time of new task is taken according to the previous task. If the server has minimum response time then it will execute the task quickly to compare with other servers.

(iv) Server Allocation

The server will get the bandwidth and minimum response time form the information server. The information taken from the information server is sorted in a particular order to find the minimum response time server. If the server has minimum response time then it will be assigned to execute the first task. The next minimum response time server will be assigned to execute the next task and so on. If the tasks are executed successfully then a mail alert will be sent to the user to grant the successful transaction.

4. Performance Analysis:

In this section, we present the performance analysis results of the FastTopK algorithm.

The correctness of the fastTopK algorithm[9] relies on the accuracy of the frequency counter information, which is the estimation of the request rates on the corresponding data items.

p(T) =total requests for T/total requests

Formally, for a data item T , we define its actual request rate FastTopK will determine an estimate

p(ˆT ) such that

p(ˆT ) ( p(T )(1 – β/2 ), p(T )(1 + /β2 ))

with probability greater than α. For example, If we set β = 0.01,

α = 99.9%, then the accuracy requirement states that with probability greater than 99.9%, FastTopK will estimate p(T ) with a relative error range of 1%. We use Zα to denote the α percentile for the unit normal distribution. For example, if α = 99.75%, then Zα = 3.

Given the specified accuracy requirement, we measure the performance of the algorithm through S and it is defined to be the number of request arrivals needed to perform the estimation. We use the term estimation time and sample size interchangeably.

We assume that request arrivals to the system follow an independent and identically distributed process. The following result is used to estimate the request rate pf from the frequency counter value of a data item f.

Let Mk(N, T) represent the frequency

counter value for the target data item T after N arrivals for fastTopK with k comparisons. We label the requests 1 to N based on the arrival sequence. Let Cij (f) = 1 both requests i and j ask for

dataitem f and 0 otherwise

Clearly, the frequency counter of data item f will be increased by 1 when

Cij (f) = 1. We call it a coincidence in the rest of

the paper.

The results follow directly from the assumption in fig.3 that arrivals are independent and that the probability that an arrival request asks for data f is pf .

We compared IWF with 3 other workload factoring algorithms:

Random: The random factoring algorithm decides with the probability D3/D1+D3 a request will go to the load group with the volume D3/D1+D3

Choke: The Choke factoring algorithm is based on the ChoKe active queue management scheme[9][13]. While ChoKe was originally proposed for approximating fair bandwidth allocation, it is a reasonable candidate for workload factoring when the optimization goal is minimizing the unique data items in one part (dropping the packets to the top popular IP destinations is similar to finding the top popular data items).

Fig.3 IWF Performance D1-Uniorm Distribution D3-Uniform Distribution RATE: The RATE factoring algorithm acts the same as IWF except that it uses the RATE scheme [14] to detect the top-K data items.

5. Experimental Results:

(5)

63

has quick response will execute the tasks first and so on. This is implemented in a banking application where the account numbers are created for the number of users and then transactions are carried out. When the transaction is successful then an alert message will be send to the user.

6. Conclusion and Future Work:

The design of a hybrid cloud computing model have proposed proactive workload management technology, the hybrid cloud computing model allows users to develop a new architecture where a dedicated resource platform runs for hosting base service workload, and a separate and shared resource platform serves flash crowd peak load[15]. Given the elastic nature of the cloud infrastructure, it creates a situation where cloud resources are used as an extension of existing infrastructures.

The response time of the server is calculated and maintained in the repository for the future allocation. The response time is the time difference between the task allocation time and task completion time. The portioned task is allocated based on the response time and the allocation is done from minimum to maximum response server time.

References:

1 Agrawal, R., and Srikant, R. “Fast algorithms for mining association rules”. In Proceedings of the Twentieth International Conference on Very Large Data Bases (1994).

2 Balachandran, A., Sekar, V., Akella, A., and Seshan,S., “Analyzing the potential benefits of CDN augmentation strategies for Internet video workloads,” 2013.

3 Casalicchio, E., Cardellini, V. and Colajanni, M., “Content-aware dispatching algorithms for cluster-based web servers,” Cluster Computing, 2002.

4 Daniel O. Awduche. MPLS and Traffic Engineering in IP Networks. IEEE Communications Magazine, December 1999.

5 Hao, F., Kodialam, M., Lakshman, T. V., and Zhang, H., “Fast payloadbased flow estimation for traffic monitoring and network security,”2005.

6 Harchol-Balter, M. , Crovella, M. E., and Murta, C. D., “On choosing atask assignment policy for a distributed server system,” 1998.

7 Jin S., and Bestavros, A., “Greedydual* web caching algorithm: exploiting the two sources of temporal locality in web request streams,” 2000.

8 Kang, X., Zhang, H., Jiang, G., Chen, H., Meng, X., and Yoshihira, K.,“Measurement, modeling, and analysis of Internet video sharing site workload: a case studyon Web Services 2008.

9 Karypis, G. , and Kumar, V. ,“Multilevel k-way hypergraph partitioning,” in Proc. 1999.

10 Kodialam, M. S. , Lakshman, T. V. , and Mohanty, S. ,“Runs based traffic estimator (rate): a simple, memory efficient scheme for per-flow rate estimation,” in 2004.

11 Lee, G. ,Patterson, D. A., Rabkin, A. ,Stoica, I., and Zaharia, M. ,“Above the clouds: a Berkeley view of cloud computing,” 2009.

12 Liu H. H, Wang Y, Yang Y. R, Wang H, and Tian C, “Optimizing cost and performance for content multihoming,” in Proc. 2012.

13 Massive (500) Internal Server Error.outage started 35 minutes ago,” Feburary 2008.

14 Pan, R. and Prabhakar, B., “CHOKe - A simple approach for providing Quality of Service through stateless approximation of fair queueing”, Sranford CSL Technical report CSL-TR-99-779, March 1999.

References

Related documents

The Active Directory PowerShell Module included in Windows Server 2012, provides over 130 cmdlets for managing Active Directory objects, such as:. • Computer Accounts • User

information will also be available on our website at www.mywealthcareonline.. up to two weeks for delivery of your card. If you do not receive your card two Inc. so that a

The synthetic data are used, firstly to learn an inverse low-dimensional to high- dimensional regression function between physical parameters and spectra from the database, and

Merchants who want to stop the ever-growing chargeback trend and protect their revenues, merchant accounts and reputation, must take steps to understand the problem. Knowing

Remember that the Outline Summary Sheet will give the examiners a framework on which to base their questions, so practise writing out the summary sheet until you feel you have a

ค าน า Preface การจัดท าสื่อ e-book เป็นการแนะน าการ พูดสื่อสารทางโทรศัพท์โดยใช้ค าส

Originality involves the exploration of the unexplored and the unanticipated (Mavodza 2010). The current research acknowledged that several studies on access to and the use

(h) Business Associate agrees to make internal practices, books, and records relating to the use and disclosure of Protected Health Information received from, or created or