Electronic Theses and Dissertation

Perkembangan teknologi informasi yang semakin pesat mendorong meningkatnya kebutuhan akan konektivitas internet yang stabil dan andal. ketergantungan pada satu penyedia layanan internet berpotensi menimbulkan permasalahan, seperti peningkatan beban trafik dan gangguan layanan (downtime), sehingga diperlukan mekanisme pengelolaan koneksi yang mampu meningkatkan keandalan dan distribusi trafik jaringan. load balancing menjadi solusi yang efektif dengan mengoptimalkan pemanfaatan koneksi dual wide area network (dual-wan) untuk mendistribusikan koneksi trafik secara proporsional. salah satu metode yang digunakan adalah per connection classifier (pcc), yang memanfaatkan fungsi hash deterministik untuk mendistribusikan trafik jaringan berdasarkan karakteristik koneksi. penelitian ini bertujuan untuk mengevaluasi kinerja load balancing pcc pada jaringan dual-wan berbasis mikrotik router operating system (routeros) dengan bandwidth asimetris, termasuk kinerja mekanisme failover, distribusi koneksi, serta utilisasi central processing unit (cpu) router. penelitian dilakukan dengan mengimplementasikan pcc pada jaringan dual-wan dengan kapasitas wan1 sebesar 20 mbps dan wan2 sebesar 3 mbps menggunakan rasio 7:1. hasil penelitian menunjukkan bahwa pcc berhasil mendistribusikan koneksi dengan rata-rata 87,7% pada wan1 dan 12,2% pada wan2, mendekati rasio teoritis 87,5% dan 12,5%. download throughput rata-rata mencapai 22,52 mbps dan upload 20,73 mbps, dengan utilisasi bandwidth sebesar 98% dan 90% dari kapasitas gabungan 23 mbps. nilai latency dan jitter pada kondisi full load masing-masing sebesar 24,2 ms dan 6,79 ms, masih termasuk kategori baik berdasarkan standar telecommunications and internet protocol harmonization over network (tiphon). utilisasi cpu meningkat hingga 47% namun masih dalam batas yang dapat diterima. mekanisme failover mencapai rata-rata downtime 6,71 detik dan failback 14,22 detik. hasil penelitian menunjukkan bahwa metode pcc mampu memanfaatkan kapasitas kedua jalur wan secara bersamaan dan menjaga ketersediaan koneksi jaringan melalui mekanisme failover yang diterapkan. kata kunci: mikortik, load balancing, per connection classifier (pcc), dual-wan, failover.

The rapid development of information technology has increased the demand for stable and reliable internet connectivity. Dependence on a single internet service provider may cause several problems, such as increased traffic load and service interruptions (downtime), making it necessary to implement a connection management mechanism that can improve network reliability and traffic distribution. Load balancing is an effective solution for optimizing the utilization of Dual Wide Area Network (Dual-WAN) connections by distributing network traffic proportionally. One of the methods used is Per Connection Classifier (PCC), which utilizes a deterministic hash function to distribute network traffic based on connection characteristics. This study aims to evaluate the performance of PCC load balancing on a dual-WAN network based on MikroTik RouterOS with asymmetric bandwidth, including the performance of failover mechanisms, connection distribution, and router Central Processing Unit (CPU) utilization. The study was conducted by implementing PCC on a dual-WAN network with WAN1 capacity of 20 Mbps and WAN2 capacity of 3 Mbps using a 7:1 ratio. The results show that PCC successfully distributed connections with an average of 87.7% on WAN1 and 12.2% on WAN2, closely matching the theoretical ratio of 87.5% and 12.5%. The average download throughput reached 22.52 Mbps and upload throughput reached 20.73 Mbps, with bandwidth utilization of 98% and 90% of the combined total capacity of 23 Mbps. The latency and jitter values under full load conditions were 24.2 ms and 6.79 ms, respectively, which are still categorized as good based on the Telecommunications and Internet Protocol Harmonization Over Network (TIPHON) standard. CPU utilization increased up to 47% but remained within an acceptable range. The failover mechanism achieved an average downtime of 6.71 seconds, while failback required 14.22 seconds. The results indicate that the PCC method is capable of utilizing the combined capacity of both WAN links simultaneously and maintaining network connection availability through the implemented failover mechanism. Keywords: MikroTik, Load Balancing, Per Connection Classifier (PCC), Dual-WAN, Failover.