Modeling and Predicting DNS Server Load

The DNS relies on caching to ensure high scalability and good performance. In optimizing caching, TTL adjustment provides a means of balancing between query load and TTLdependent performances such as data consistency, load balancing, migration time, etc. To gain the desired balance, TTL adjustment depends on predictions of query loads under alternative TTLs. This paper proposes a model of DNS server load, which employs the uniform aggregate caching model to simplify the complexity of modeling clients’ requests and their caching. A method of predicting DNS server load is developed using that model. The prediction method is solely based on the unilateral measurements or observations at authoritative servers. Without reliance on lots of multi-point measurements nor distributed measuring facilities, the method is best suited for DNS authoritative operators. The proposed model and prediction method are validated through extensive simulations. Finally, global sensibility analysis is conducted to evaluate the impacts of measurement uncertainties or errors on the predictions.