LOAD BALANCING USING LSTM NETWORK AND DOCKER

Document Type : Primary Research paper

Authors

1 Department of Computer Science and Engineering, SRM Institute of Science and Technology, Kattankulathur, Chennai, India

2 Department of Computer Science and Engineering, Vel Tech Rangarajan Dr.Sagunthala R&D Institute of Science and Technology, Chennai, India

Abstract

– Recurrent Neural Network is widely used in Natural Language processing (NLP) based task like Automatic Speech Recognition(ASR), Speaker Gender Identification, Speaker Identification, Speaker Emotion recognition. It is proved that RNN works well in time series data and also provide better accuracy in above mentioned Research task. In this proposed work we have implemented Long short-term memory (LSTM) a type of RNN for Load balancing in web server. Storing information and providing service to client based on their request is the primary work of web server. One of the major issues in web server is Load balancing. Existing methodologies for load balancing usually depend upon both hand-crafted infrastructure scales up and/or rule based algorithms such as scaling up when the server loads (CPU/memory) hit high enough to trigger rule based load balancing servers or by observing the requests manually and scaling up as needed. These techniques can result in significant delays during the rush hour. Our methods provide a supervised learning methodology. A recurrent neural network was devised to predict loads on the servers. The proposed system utilized a LSTM network to predict the said loads. The data from the LSTM network will be used to create additional containers (docker containers) to handle the load before it even happens hence preventing the system from any down time.

Keywords

Int. J. of Aquatic Science
Volume 12, Issue 2 - Serial Number 2
June 2021
Pages 2205-2213

Files

Share

How to cite

Statistics