Analisis Kinerja Manajemen Disk SSD Berdasarkan Pola Akses Data dan Spesifikasi Perangkat Menggunakan CrystalDiskMark

Abstract

The rapid development of computing systems in the modern digital era has increased the demand for high-performance and reliable storage devices. Solid State Drives (SSDs) are widely adopted due to their superior speed and efficiency; however, their actual performance may vary depending on data access patterns and hardware specifications. This study aims to analyze the performance of SSD disk management based on different data access patterns and device specifications using CrystalDiskMark. This research employs a quantitative experimental approach involving 16 laptops with varying RAM capacities (2–32 GB) and processor types ranging from entry-level to high-performance architectures. Performance testing was conducted under two main scenarios: (1) read/write operations on a large single file (4 GB) representing sequential access, and (2) read/write operations on a folder containing 2000 small files representing random access. CrystalDiskMark was used as the benchmarking tool, and each test was repeated three times to ensure result consistency. The collected data were analyzed using a comparative descriptive method. The results show that SSD performance is significantly influenced by data access patterns and hardware configurations. Sequential access to large files produces higher and more stable read/write speeds compared to random access involving small files, which results in substantial performance degradation across all devices. Devices equipped with larger RAM capacities and modern processors demonstrate better stability and higher throughput, particularly under random workloads. Conversely, systems with limited RAM and entry-level processors experience severe performance bottlenecks despite using SSDs. This study concludes that SSD performance is not solely determined by the storage medium itself but is strongly affected by the synergy between data access patterns, memory capacity, and processor capability. The findings provide empirical evidence to support optimal hardware configuration selection for different computing workloads.