Macroscopic Workload Management and Mainframe Capacity Planning

In: Tagungsband Vierter Workshop des GI-ITG-Fachausschusses MMB zur Leistungs-, Zuverlässigkeits- und Verlässlichkeitsbewertung von Kommunikationsnetzen und verteilten Systemen / Heidtmann, K.; Wolfinger, B. (Hrsg.)
Hamburg (2007)
Buchaufsatz / Kapitel / Fach: Informatik; Wirtschaftswissenschaften
Since in capacity planning activities the available information with respect to workload
and system configuration is often rather coarse we propose a technique which is
based on a “macroscopic” (high level) view. Based on the analysis of utilization data
collected from mainframe monitoring we introduce a heuristic algorithm for the
allocation of logical partitions (LPARs) to multiprocessors. Subsequently queuing
theoretic formulas for M/M/m- and G/G/m-models can be used to check to which
extent a proposed LPAR allocation will fulfil prespecified service levels. The approach
is illustrated by data taken from real world scenarios.