Biased distribution; Non-uniform distribution
Data skew primarily refers to a non uniform distribution in a dataset. Skewed distribution can follow common distributions (e.g., Zipfian, Gaussian, Poisson), but many studies consider Zipfian  distribution to model skewed datasets. Using a real bibliographic database,  provides real-world parameters for the Zipf distribution model. The direct impact of data skew on parallel execution of complex database queries is a poor load balancing leading to high response time.
Walton et al.  classify the effects of skewed data distribution on a parallel execution, distinguishing intrinsic skew from partition skew. Intrinsic skew is skew inherent in the dataset (e.g., there are more citizens in Paris than in Waterloo) and is thus called Attribute value skew (AVS). Partition skew occurs on parallel implementations when the workload is not evenly distributed between nodes, even when input data is uniformly...
- 1.Zipf GK. Human behavior and the principle of least effort: an introduction to human ecology. Reading: Addison-Wesley; 1949.Google Scholar
- 2.Lynch C. Selectivity estimation and query optimization in large databases with highly skewed distributions of column values. In: Proceedings of the 14th International Conference on Very Large Data Bases; 1988. p. 240–51.Google Scholar
- 3.Walton CB, Dale AG, Jenevin RM. A taxonomy and performance model of data skew effects in parallel joins. In: Proceedings of the 17th International Conference on Very Large Data Bases; 1991. p. 537–48.Google Scholar