Analisis Performa Sistem Grab POI dalam Pengelolaan Big Data Geospasial Menggunakan Pendekatan Real-Time Processing

Sari

Abstrak: Perkembangan teknologi digital pada era Industri 4.0 memfasilitasi penerapan data geospasial dalam aplikasi Location-Based Services (LBS), termasuk platform Grab yang mengelola big data Point of Interest (POI). Tantangan utama mencakup latensi yang tinggi, konsistensi data, serta kompleksitas struktur data geospasial yang menuntut pemrosesan secara real-time. Kajian ini mengevaluasi kinerja sistem POI Grab melalui pendekatan eksperimental kuantitatif, dengan memanfaatkan arsitektur pipeline real-time yang melibatkan komponen seperti Apache Kafka, Flink/Spark Streaming, dan penyimpanan terdistribusi. Metodologi mencakup simulasi data POI berskala besar, pengukuran indikator kinerja seperti latensi, throughput, dan akurasi, serta analisis variabel seperti volume data dan mekanisme indeksasi. Temuan menunjukkan pengurangan latensi sebesar 33%, peningkatan throughput hingga 20%, dan akurasi mencapai 95%, yang mengonfirmasi keefektifan pemrosesan real-time dalam mendukung layanan Grab. Saran untuk pengembangan lebih lanjut meliputi optimasi indeksasi geospasial dan implementasi auto-scaling guna meningkatkan skalabilitas.

Kata kunci: big data geospasial, pemrosesan real-time, Point of Interest (POI), Grab, Location-Based Services

Abstract: The evolution of digital technology in the Industry 4.0 era facilitates the application of geospatial data within Location-Based Services (LBS), exemplified by Grab's platform managing extensive Point of Interest (POI) big data. Primary challenges encompass elevated latency, data consistency issues, and the intricate nature of geospatial data structures necessitating real-time processing. This research evaluates the performance of Grab's POI system via a quantitative experimental methodology, utilizing a real-time pipeline architecture incorporating components such as Apache Kafka, Flink/Spark Streaming, and distributed storage. The approach involves large-scale POI data simulation, measurement of performance metrics including latency, throughput, and accuracy, alongside assessment of factors like data volume and indexing mechanisms. Findings reveal a 33% reduction in latency, a 20% enhancement in throughput, and 95% accuracy, validating the efficacy of real-time processing in supporting Grab's services. Recommendations for further development include geospatial indexing optimization and auto-scaling implementation to bolster scalability.

Keywords: geospatial big data, real-time processing, Point of Interest (POI), Grab, Location-Based Services