Scaling Control Tra c in Network Protocols

The primary goal of a data network is to carry data tra c. Some of the tra c carried in a network is control tra c such as routing, signalling and end-to-end protocol control. Unregulated growth of control tra c can jeopardize the primary goal of networks. This paper discusses scaling of control tra c in network protocols and proposes some general scaling techniques. Control tra c can be regulated along three scaling dimensions: (1) frequency, (2) distribution scope, and (3) information aggregation. This paper describes a study of scalable timers to regulate control tra c along frequency dimension. We also describe future work in the areas of distribution scope and information aggregation. Soft state protocols use periodic refresh messages to keep network state alive while adapting to changing network conditions; this has raised concerns regarding the scalability of protocols that use the soft-state approach. In existing soft state protocols, the values of the timers that control the sending of these messages, and the timers for aging out state, are chosen by matching empirical observations with desired recovery and response times. These xed timer-values fail because they use time as a metric for bandwidth; they adapt neither to (1) the wide range of link speeds that exist in most wide-area internets, nor to (2) uctuations in the amount of network state over time. We propose and evaluate a new approach scalable timers in which timer-values adapt dynamically to the volume of control tra c and available bandwidth on the link. Global distribution of multi-party control messages in existing protocols can lead to unconstrained growth of control tra c. Aggregation ( or merging) of control messages and then hierarchical distribution can provide better scaling properties. In this paper we propose various techniques that can be used to distribute control messages for group state synchronization with limited scope.