ENERGY-AWARE LOAD BALANCING AND APPLICATION SCALING FOR THE CLOUD ECOSYSTEM

ENERGY-AWARE LOAD BALANCING AND APPLICATION SCALING FOR THE CLOUD ECOSYSTEM

ABSTRACT:

In this paper we introduce an energy aware operation model used for load balancing and application scaling on a cloud. The basic philosophy of our approach is defining an energy optimal operation regime and attempting to maximize the number of servers operating in this regime. Idle and lightly –loaded servers are switched to one of the sleep states to save energy. The load balancing and scaling algorithms also exploit some of the most desirable features of server consolidation mechanisms discussed in the literature.

EXISTING SYSTEM:

In the last few years packaging computing cycles and storage and offering them as a metered service became a reality. Large farms of computing and storage platforms have been assembled and a fair number of cloud service providers(CSPs) offer computing services based on three cloud delivery models SAAS (software as a service).PAAS (Platform As A Service) and IASS(Infrastructure as a service).Warehouse scale computers(WSCS) are the building blocks of a cloud infrastructure. A hierarchy of networks connects 50; 000 to 100; 000 servers in a WSC. The servers are housed in racks; typically, the 48 servers in a rack are connected by a 48-port Gigabit Ethernet switch. The switch has two to eight up links which go to the higher level switches in the network hierarchy .Cloud elasticity the ability to use as many resources as needed at any given time, and low cost, a user is charged only for the a resources it consumes, represents solid incentives for many organizations to migrate their computational activities to a public cloud.

The number of CSPs, the spectrum of services offered by the CSPs, and the number of cloud users have increased drastically during the last few years. For example, in 2007 the EC2 (Elastic Computing Cloud) was the first service provided by AWS (Amazon Web Services); five year later, in 2012, AWS was used by businesses in 200 countries. Amazon’s S3 (Simple Storage Services) has surpassed two trillion objects and routinely runs more than 1.1 million peak requests per second.

PROPOSED SYSTEM:

There are three primary contributions of this paper

(1) a new model of cloud servers that is based on different operating regimes with various degrees of energy efficiency” (processing power versus energy consumption);

(2) a novel algorithm that performs load balancing and application scaling to maximize the number of servers operating in the energy-optimal regime; and

(3) analysis and comparison of techniques for load balancing and application scaling using three differently-sized clusters and two different average load profiles.

 Models for energy-aware resource management and application placement policies and the mechanisms to enforce these policies such as the ones introduced in this paper c an be evaluated theoretically, experimentally, though simulation, based on published data, or though a combination of these techniques. Analytical models can be used to derive high-level insight on th behavior of the system in a very short time but the biggest challenge is in determining the values of the parameters; while the results from an analytical model can give a good approximation of the relative trends to expect, her may be significant errors in the absolute predictions.

Experimental data is collected on small scale systems such experiments provide useful performance data for individual system components but no insights on the interaction between the system and applications and the scalability of the policies. Trace based workload analysis such as the ones are very useful though they provide information for a particular experiment set up hard ware configuration and applications. Typically trace based simulation need more time to produce results.

 Traces can also be very large and it is hard to generate resprestative traces from one class of machines that will be valid for all the classes of simulated machines. To evaluate the energy aware load balancing and application scaling policies and mechanism introduced in this paper we chose simulation using data published in the literature.

MODULE 1

LOAD BALANCING IN CLOUD COMPUTING

Cloud computing” is a term, which involves virtualization, distributed computing, networking, software and web services. A cloud consists of several elements such as clients, datacenter and distributed servers. It includes fault tolerance, high availability, scalability, flexibility, reduced overhead for users, reduced cost of ownership, on demand services etc. Central to these issues lies the establishment of an effective load balancing algorithm. The load can be CPU load, memory capacity, delay or network load. Load balancing is the process of distributing the load among various nodes of a distributed system to improve both resources utilization and job response time while also avoiding a situation where some of the nodes are heavily loaded while mother nodes are idle or doing very little work. Load balancing ensures that all the processor in the system or every node in the network does approximately the equal amount of work at any instant of time. This technique can be sender initiated, receiver initiated or symmetric type (combination of sender initiated and receiver initiated types). Our objective is to develop an effective load balancing algorithm using Divisible load scheduling theor to maximize or minimize different performance parameters (throughput, latency for example) for the clouds of different sizes ( virtual topology depending on the application requirement).

MODULE 2

ENERGY EFFICIENCY OF A SYSTEM

The energy efficiency of a system is captured by the ratio performance per Watt of power.” During the last two decades the performance of computing systems has increased much fater than their energy efficiency. Energy proportional systems. In an ideal world, the energy consumed by an idle system should be near zero and grow linearly with the system load. In real life, even systems whose energy requirements scale linearly,  when idle, use more than half the energy they use at full load. Data collected over a long period of time shows that the typical operating regime for data center servers is far from an optimal energy consumption regime. An energy-proportional system consumes no energy when idle, very little energy under a light load, and gradually, more energy as the load increases. An ideal energy proportional system consumes no energy when idle, very little energy uder a light load and gradually more energy as the load increases. An ideal energy proportional system is always operating at 100% efficiency.

 Energy efficiency of a data center the dynamic range of subsystems. The energy efficiency  of a data center is measured by the power  usage effectiveness(PUE), the ratio of total energy used to power a data center to the energy used to power computational servers, storage servers, and other IT equipment. The PUE has improved from around 1:93 in 2003 to 1:63 in 2005 recently, google reported a PUE ratio as low as 1.15. the improvement in PUE forces us to concentrate on energy efficiency of computational resources. The dynamic range is the difference between the upper and the lower limits of the energy consumption of a system as a function of the load placed on the system. A large dynamic range means that a system is able to operate at a lower fraction of its peak energy when its load is low. Differnet subsystems of a computing systems behave differently in terms of energy efficiency while many processors have reasonably good energy proportional profiles significant improvements in memory and disk subsystems are necessary. The largest consumer of energy in a server is the processor, followed by memory, and storage systems. Estimated distribution of the peak power of different hardware system in one of the Google’s datacenters is CPU 33%, DRAM 30%.

Disks 10%, network 5%, and others 22%. The power consumption can vary from  45w to 200w per multicore CPU. The power consumption of servers has increased over time during the period 2001-2005 the estimated average power se has increased from 193 to 225 W for volume servers, from 457 to 675 for mid range severs and from 5,832 to 8,163 W for high end ones. Volume servers have a price less than $25 K, mid-range servers have a price between $25 K and $499 K, and high-end servers have a price tag larger than $500 K. Newer processors include power saving technologies. The processors used in servers consume less than one third of their peak power at very-low load and have a dynamic range of more than 70% of peak power the processors used in mobile and /or embedded applications are better in this respect. According to the dynamic power range of other components of a system is much narrower less than 50% for DRAM, 25% for disk drivers, and 15% for networking switches. Larger servers often use 32 64 dual in line memory modules (DIMMs) the power consumption of one DIMM is in the 5 to 21 W range.

MODULE 3

RESOURCE MANAGEMENT POLICIES FOR LARGER SCALE DATA CENTERS.

These policies can be loosely grouped into five classes:

·         Admission control

·         Capacity allocation

·         Load balancing

·         Energy optimization

·         Quality of service(QOS) guarantees.

The explicit it goal of an admission control policy is to prevent the system from accepting workload in violation of high level system policies a system should not accept additional workload preventing it from completing work already in progress or contracted. Limiting the workload requires some knowledge of the global state of the system; in a dynamic system such knowledge, when available, is at best obsolete. Capacity allocation means allocating resources for individual instances; an instance is an activation of a service. Some of the mechanisms for capacity allocation are based on either static or dynamic thresholds. Economy of scale aspects the energy efficiency of data processing.

For example, google reports that the annual energy consumption for an Email service varies significantly depending on the business size and can be 15 times larger for a small business than for a larger one. Cloud computing can be more enrgy efficient than on premise computing for many organizations.

MODULE 4

SERVER CONSOLIDATION

·         The term server consolidation is used to describe:

·         Switching idle and lightly loaded systems to a sleep state

·         Workload migration to prevent overloading of systems

Any optimization of cloud performance and energy efficiency by redistributing the workload.

Server consolidation policies. Several policies have been proposed to decide when to switch a server to a sleep state. The reactive policy responds to the current load; it switches the servers to a sleep state when the load decreases and switches them to the running state when the load increases. Generally, this policy leads to SLA violations and could work only for slow varying predictable loads. To reduce SLA violations one can envision a reactive with extra capacity policy when one attempts to have  a safety margin and keep running a fraction of the total number of servers, e.g., 20% above those needed for the current load. The auto scale policy is a very conservative reactive policy in switching servers to sleep state to avoid the power consumption and the delay in switching them back to running state. This can be advantageous for un predictable policies. The moving window policy estimates the workload by measuring the average request rate in a window of size seconds uses this average to predict the load for second(_+2). And so on. The predictive linear regression policy uses a linear regression to predict the future load.

MODULE 5

ENERGY AWARE SCALING ALGORITHMS

The objective of the algorithms is to ensure that the large possible number of active servers operate within the boundaries of their respective optimal operating regime. The actions implementing this policy are:

·         Migrate VMs from a server operating in the undesirable low regime and then switch the server to a sleep state;

·         Switch an idle server to a sleep state and reactive servers in a sleep state when the cluster load increases

·         Migrate the VMs from an overloaded server, a server operating in the undesirable high regime with applications predicated to increase their demands for computing in the next teallocation cycles.

For example, when deciding to migrate some of the VMs running on a server or to switch a server to a sleep state, we can adopt a conservative policy similar to the one advocated by auto scaling to save energy.Predicitve policies, such as the ones discussed will be used to allow a server to operate in a suboptimal regime when historical data regarding its workload indicates that it is likely to return to the optima regime in the near future.

The cluster leader has relatively accurate information about the cluster load and its trends. The leader could use predictive algorithms to initiate a gradual wake up process for servers in a deeper sleep state, C4 C6, when the workload is above a high water mark and the workload is continually increasing. We set up the high water mark at 80% of the capacity of active servers; a threshold of 85% s used for deciding that a server is overloaded, based on an analysis of workload traces. The leader could also choose to keep a number of servers in C1 or C2 states because it takes less energy and time to return to the C0 state from these states. The energy management component of the hypervisor can use only local information to determine the regime of a server.

CONCLUSION:

The realization that power consumption of cloud computing centers is significant and is expected to increase substantially in the future motivates the inters of the research community in energy aware resource management and application placement policies and the mechanisms to enforce these policies. Low average server utilization and its impact on the environment make it imperative to devise new energy aware policies which identify optimal regimes for the cloud servers and at the same time, prevent SLA violations. A Quantitative evaluation of an optimization algorithm or an architectural enhancement is a rather intricate and time consuming process; several benchmarks and system configurations are used to gather the data necessary to guide future developments.

For example, to evaluate the effects of architectural enhancements supporting instruction level or data level Parallelism on th processor performance and their power consumption several benchmarks are used. The results show numerical outcomes for the individual applications in each benchmark.

Similarly, the effects of an energy aware algorithm depend on the system configuration and on the application and cannot be expressed by a single numerical value. Research on energy aware resource management in large scale numerical value.

Research on energy aware resource management in large scale systems often use simulation for a quasi quantitative and more often a qualitative evaluation of optimization algorithms of procedures.

As stated Firs,they(WSCs) are a new class of large-scale machines driven by a new and rapidly evolving set of workloads. Their size alone makes them difficult to experiment with or to simulate efficiently. It is rather difficult to experiments with the systems discussed in this paper and this precisely the reason why we choose simulation.

Author(s)

Paya,A. Ashkan Paya is with Computer Science Division, EECS Department University of Central Florida, Orlando, FL 32816, USA. (email:ashkan paya@knights.ucf.edu)
Marinescu, D.