Holistic Load-Balancing in a Distributed Spatial Cache

Abstract

A steadily growing number of people using location based services (LBS) inflict massive query loads on the data tier of an LBS. As such queries usually possess considerable overlap, multiple cache nodes collaborating in a distributed spatial cache can provide scalable access to frequently used data. To preserve high throughput throughout the complete execution process, it is necessary to balance the accumulating load among the participating cache nodes. In this work, we identify three key-indicators to improve resource utilization during the load-balancing process: data skew, anticipated data access patterns and dynamic load peaks. For this reason, we introduce a comprehensive mathematical model to express the key-indicators as probability distribution functions. We fuse the different key-indicators into a single holistic distribution model. In the course of this, we devise a methodology from our holistic distribution model towards a distributed spatial cache offering improved load-balancing facilities.

An Elastic Cache Infrastructure through Multi-level Load-balancing

“…As this paper builds upon the ideas of previous work (Lübbe et al, 2011;Lübbe et al, 2012;Lübbe et al, 2013), this section outlines their major contributions. First, we describe the general architecture of the distributed main memory cache.…”

“…Abstract Distribution Model. In (Lübbe et al, 2013), we introduced an abstract mathematical model to describe load distributions as probability density functions. Formally, such a distribution function can be defined as a two-dimensional function ρ : R 2 → R which maps geographic coordinates to an abstract measure representing load.…”

“…This overcompensation can lead to poor resource utilization which compromises the overall performance of the system, as the resources would have been needed elsewhere, e.g., to cache data of insular regions. The different load-balancing indicators data skew, expected workload and dynamic load peaks can all be consolidated into a comprehensive mathematical model (Lübbe et al, 2013). In this work, we implement a distributed and elastic main memory cache supporting multi-level load-balancing which considers the different indicators of the abstract model in separate stages.…”