Service level measurement based on Uptime data monitoring for rural internet access services in Indonesia

摘要

As an archipelago country, Indonesia still has many islands which do not have proper infrastructure. This situation creates a huge digital divide which made rural areas often are lagging behind from the cities in terms of economy and public health. To bridge this gap, a government organization initiated a program to provide internet access in rural areas, which then joined by many ISPs. Problem arose when initial method to calculate service level, which based on Ping sensor downtime, cannot be implemented. While the method can detect a failure, it cannot identify a failure which caused by power failure or by transmission (link) failure, so the service level will be significantly lower than it should be. Whereas, the service contract clearly says that the failure which caused by power cannot be charged to ISPs (tolerance factor). To accommodate the contract, a new method to calculate service level shall be made. This paper proposes a new method based on Uptime sensor downtime, which can be collected from Network Management System (NMS). The new method observes the duration of a device's uptime over time and gives different and unique visualization for each type of failure, transmission or power failure. This allow us to calculate service level objectively and effectively. The method gives remarkably improvement in terms of service level percentage. As a comparison, in a simulation test a new method gave 97,5% service level while the old method which based on Ping downtime gave 56% of service level. The huge difference is attributed to many power failures which cannot be identified by the old method and eventually cannot be emitted from the service level calculation.