On Ranges and Partitions in Optimal TCAMs

TL;DR

This paper investigates how to efficiently represent traffic splitting partitions in TCAMs, revealing trade-offs between memory minimization and fragmentation, especially for range-based functions with contiguous address segments.

Contribution

It introduces a novel analysis of the trade-offs between minimal TCAM representations and fragmentation in range-based traffic splitting functions.

Findings

Minimizing TCAM entries can cause fragmentation in partitions.

Some actions require multiple ranges in minimal representations.

Single-segment constraints for each target increase representation complexity.

Abstract

Traffic splitting is a required functionality in networks, for example for load balancing over paths or servers, or by the source's access restrictions. The capacities of the servers (or the number of users with particular access restrictions) determine the sizes of the parts into which traffic should be split. A recent approach implements traffic splitting within the ternary content addressable memory (TCAM), which is often available in switches. It is important to reduce the amount of memory allocated for this task since TCAMs are power consuming and are often also required for other tasks such as classification and routing. In the longest-prefix model (LPM), Draves et al. (INFOCOM 1999) find a minimal representation of a function, and Sadeh et al. (INFOCOM 2019) find a minimal representation of a partition. In certain situations, range-functions are of special interest, that is, all the addresses with the same target, or action, are consecutive. In this paper we show that minimizing the amount of TCAM entries to represent a partition comes at the cost of fragmentation, such that for some partitions some actions must have multiple ranges. Then, we also study the case where each target must have a single segment of addresses.