Task programming in cloud computing
0 likes475 views
This document discusses task programming in cloud computing. It introduces task computing as a way to distribute code execution across remote computing nodes. The document then describes an experiment using RabbitMQ to distribute data analysis tasks across varying numbers of workers. The results show that increasing the number of workers reduces the total time to complete the tasks. Finally, the document concludes that task programming is well-suited for applications that require large-scale distributed computing over long periods of time and can provide a low-cost alternative to grid computing through the use of cloud resources.
![References
[1] Lee, Edward A. "The problem with threads." Computer 39.5 (2006): 33-42.
[2] Buyya, Rajkumar, Christian Vecchiola, and S. Thamarai Selvi. Mastering cloud computing: foundations and
applications programming. Newnes, 2013.
[3] Gunarathne, Thilina, et al. "MapReduce in the Clouds for Science." 2010 IEEE second international conference
on cloud computing technology and science. IEEE, 2010.
[4] Ge, Yujia, and Guiyi Wei. "GA-based task scheduler for the cloud computing systems." 2010 International
Conference on Web Information Systems and Mining. Vol. 2. IEEE, 2010.
[5] Lin, Weiwei, et al. "Bandwidth‐aware divisible task scheduling for cloud computing." Software: Practice and
Experience 44.2 (2014): 163-174.
[6] Mocanu, Eleonora Maria, et al. "Cloud computing—task scheduling based on genetic algorithms." 2012 IEEE
International Systems Conference SysCon 2012. IEEE, 2012.
[7] Rostanski, Maciej, Krzysztof Grochla, and Aleksander Seman. "Evaluation of highly available and fault-tolerant
middleware clustered architectures using RabbitMQ." 2014 federated conference on computer science and
information systems. IEEE, 2014.
[8] Foster, Ian, et al. "Cloud computing and grid computing 360-degree compared." arXiv preprint arXiv:0901.0131
(2008).](https://image.slidesharecdn.com/taskprogrammingincloudcomputing-210513204504/85/Task-programming-in-cloud-computing-10-320.jpg)
Task programming in cloud computing
- 1. Task Programming in Cloud Computing Presented By: Suresh Pokharel (074MSCSK015) M.Sc. in Computer System and Knowledge Engineering Institute of Engineering, Pulchowk Campus
- 2. Introduction : Task Computing ● A distinct unit of code, or a program, that can be separated and executed in a remote runtime environment ● Task computing provides distribution by harnessing the computer power of several computing nodes ● Task computing is categorized into: High-performance computing (HPC), High-throughput computing (HTC) and Many-task computing (MTC).
- 3. Introduction ● Used a RabbitMQ, A popular open source message broker to distribute a task on queue ○ Asynchronous Messaging system ○ 100% Availability ○ Distributed Deployment ○ Enterprise & Cloud Ready ● A large task was divided into different independent subtask and supplied to multiple worker to perform the task
- 4. Methodology Data Set: ● Numeric value in large volume ● an individual household electric power consumption time series data set with 2,075,259 different records are considered, each record consists of 9 attributes but this research has considered only 4 attributes
- 5. Methodology(Cont...) Task Separation and Distribution: ● The objective of the workers is to find average of Sub_metering_1, Sub_metering_2, Sub_metering_3. ● To accomplish the task, 2075259 records are divided into different 1442 sub-tasks grouped by date attribute ● Message queueing system used to queue the task for higher availability to workers ● Available workers take the task complete them and wait for next task to be assigned by message broker
- 6. Experiment ● Cloud Platform Used: Heroku ● Server : ○ Memory (RAM): 512 MB ○ Processor: 4 virtual core Intel Xeon X5550 @ 2.67GHz ○ CPU Share: 1x ○ OS: Ubuntu 18.04 LTS ● Internet bandwidth: 5 Mbps (Approx.) ● The main task is performed by varying number of workers i.e. 1, 2, 3, 4 and 5 workers and each worker with Intel core i3-8145U, 3.45 GHz Processor and 4GiB main memory
- 7. Result ● The time taken to complete the task using different configuration is as follows: MapReduce Programming Model: Machine configurations: I5-4200U intel core 1.6-2.3 ghz(4 cpu) Main Memory of 4 GiB and result is completed in 3.6s
- 9. ● Task based programming is most intuitive approach for distributing the computation of an application over set of nodes ● Increasing number of workers reduces the total work completion time ● Data to compute in task programming must have large computing power over a long period of time ● By use of cloud computing it can be low cost alternative over grid computing Conclusion
- 10. References [1] Lee, Edward A. "The problem with threads." Computer 39.5 (2006): 33-42. [2] Buyya, Rajkumar, Christian Vecchiola, and S. Thamarai Selvi. Mastering cloud computing: foundations and applications programming. Newnes, 2013. [3] Gunarathne, Thilina, et al. "MapReduce in the Clouds for Science." 2010 IEEE second international conference on cloud computing technology and science. IEEE, 2010. [4] Ge, Yujia, and Guiyi Wei. "GA-based task scheduler for the cloud computing systems." 2010 International Conference on Web Information Systems and Mining. Vol. 2. IEEE, 2010. [5] Lin, Weiwei, et al. "Bandwidth‐aware divisible task scheduling for cloud computing." Software: Practice and Experience 44.2 (2014): 163-174. [6] Mocanu, Eleonora Maria, et al. "Cloud computing—task scheduling based on genetic algorithms." 2012 IEEE International Systems Conference SysCon 2012. IEEE, 2012. [7] Rostanski, Maciej, Krzysztof Grochla, and Aleksander Seman. "Evaluation of highly available and fault-tolerant middleware clustered architectures using RabbitMQ." 2014 federated conference on computer science and information systems. IEEE, 2014. [8] Foster, Ian, et al. "Cloud computing and grid computing 360-degree compared." arXiv preprint arXiv:0901.0131 (2008).
- 11. Thank You!