KINERJA HYBRID MONGODB-ELASTICSEARCH PADA APLIKASI SOCIAL NETWORK ANALYSIS

Abstract

Social media data is growing rapidly in size, variety and complexity. Social media data stores various potential information such as sentiment analysis, trend predictions etc. Potential information can be extracted through Social Network Analysis. SNA has a major challenge which is to process very large datasets in a reasonable time. One of the efforts that can be done is create hybrid system of MongoDB and Elasticsearch using social media datasets from Twitter. The results of this study that the highest response time in insert process starting from 26.2s on 1K data to 19520.45s on 1M data. The replication process with 1K tweet data is 6.25s to 1M tweets is 2817,146s. The select process has under 0.1s and relatively constant due to the Inverted Index on Elasticsearch. Highest CPU performance in process of selecting data from Elasticsearch. Highest RAM performance in the insert process to MongoDB and data replication to Elasticsearch.