Author: Mustafa Ali AKCA, Tuncay AYDOĞAN
Publishing Date: 2016
Volume: 4 Issue: 2
Huge amount of data stack which cannot be stored nor processed by traditional methods is called Big Data. This term, which is becoming more and more popular, led to the necessity of tools to process this data. One of the tools which is used for analysis and storage of this huge among of data is Elasticsearch. Elasticsearch is a content analysis and search server based on Lucene and developed in Java as open source. It is a software which can operate as distributed architectural structure. It also can store data in different shards in the same index, in different files in the same disk, in different disks in the same computer, and in different servers in the same network. All these options are shaped by the needs of users. As soon as Elasticsearch node starts working, it takes an active role in all indexes of clusters. It also connects with other nodes and the share of load takes place. This load distribution normally aims to increase performance by decreasing load in each node. However , this load distribution done automatically by Elasticsearch might not always create effects which increase performance. With the software developed in this study, load in each Elasticsearch nodes are tracked and manual configuration is enabled. This software enables users to observe node activity rates, to distribute shards in indexes manually, to switch on and off shard automatically, to index all these configuration productivity, and to test as inquiry-based.
Key Words: Elasticsearch, Load Distribution, Load Balancing