International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395 -0056
Volume: 04 Issue: 02 | Feb -2017
p-ISSN: 2395-0072
www.irjet.net
Efficient Dynamic Load-Balance Flow Scheduling in cloud for Big Data Centers. Snehal Ghodake1, Sulochana Sonkamble2 1Department of Computer Engineering Rajarshi Shahu School of Engineering and Research
Pune ,Maharashtra, India. 2Dr,Department
of Computer Engineering Rajarshi Shahu School of Engineering and Research Pune ,Maharashtra, India. ---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - Big data centers in cloud, large amount of data
Traditional hardware-based load balancing techniques cannot be widely used due to the high cost and the deficiency in programmable ability. Therefore, more and more researchers pay more attention on software-defined networking(SDN)techniques (e.g.,OpenFlow) that can improve transmission capacity of data centers through programmable load balanced flow control [1]. Many schemes have been proposed for load-balanced flow scheduling in OpenFlow based networks [8] [10]. They focus on the initial route selection only before the flow transmission. Network states and work load, however, often dynamically change because during a data transmission, a part of links may become unavailable, new data flows can arrive and some existing data flows have completed . As a result, the existing proposals cannot meet the needs of dynamical load balance during data migrations. On the other hand, as data center networks become more large and more complex, the time that these proposals need for the initial path selection will increase tremendously [4] [5]. Motivated by the above observations, we in this paper propose a efficient novel dynamical load-balanced scheduling (DLBS) approach to maximize the network throughput through dynamically balancing data flows. Aiming at the most popular two OpenFlow network models, three-layer non-blocking fully populated network (FPN) and three-layer fat-tree network (FTN), we will propose and develop different scheduling algorithms, which quantitatively analyze the imbalance degree of data center networks at the beginning of each time slot and then schedule unbalanced data flows once a load imbalance happens. Traditional OpenFlow framework has a few limitations [10], for example, it does not support hash-based routing [9] to spread traffic over multiple paths. So, we rely more on flexible load balancing algorithms to fit different imbalance situations. To the best of our knowledge, this is the first work to switch flows in the midway of data transmissions in OpenFlow networks. The main contributions of this paper are summarized as follows.
needs to be transferred frequently among thousands of interconnected servers. In which load balancing and flow scheduling is a challenging issue. The Open Flow is a auspicious solution to balance data flows in big data center network through programmatic traffic controller. Existing solution can able to statically set up routes only at initialization stage of data transmissions, which experiences from dynamical flow distribution and network changing state so it results in decrease system performance. In this paper, we will propose a new dynamical load-balanced scheduling (DLBS) approach for increase the network throughput to dynamically balance workload. This approach formulate the DLBS problem, and then develop a set of improved heuristic scheduling algorithms for the two typical Open Flow network models, which balance data flows time slot by time slot. Experimental results demonstrate that our DLBS approach signicantly outperforms other load-balanced scheduling algorithms Round Robin and LOBUS; and the higher imbalance degree data flows in data centers exhibit, the more improvement our DLBS approach will bring to the data centers. Key Words: Big data centre, Open Flow, Dynamical load balanced scheduling.
1.INTRODUCTION In computing, the distribution of workloads across multiple computing resources, such as com-puters, a computer cluster, network links, central processing units, or disk drives. Load balancing aims to optimize resource use minimize response time, maximize throughput and avoid overload of any single resource. Using multiple components with load balancing instead of a single component may increase reliability and availability through redundancy. Load balancing usually involves dedicated software or hardware, such as a multilayer switch or a Domain Name System server process.
Š 2017, IRJET
|
Impact Factor value: 5.181
|
ISO 9001:2008 Certified Journal
|
Page 1245