Dynamic Resource Adjustment for Performance Guarantee of IoT Directory Service

摘要

Since the current DNS/LDAP systems are unable to meet the scalability and latency requirements of IoT services, we have been investigating a new IoT directory service, which should be capable to store a huge number of records containing various attributes and perform fast lookup and update. It is constructed on virtualized network and computing resources so that the amount of resources assigned to it can be elastically adjusted to maintain the performance at desired levels despite fluctuation in workload. We present a mechanism for dynamic control of resources by regularly monitoring the performance and resource utilization. This mechanism is being implemented with Docker containers. The preliminary results to validate the concept of maintaining the latency performance in a desired range by dynamic control of CPU resource are also presented.