Synthetic workload generation for capacity planning of virtual server environments Hugo E. S. Galindo, Wagner M. Santos, Paulo R. M. Maciel, Bruno Silva and S´ ergio M. L. Galdino Centro de Inform´ atica Universidade Federal de Pernambuco Recife, Pernambuco, Brasil Jos´ e Paulo Pires Hewlett-Packard (HP) Brasil Porto Alegre, Rio Grande do Sul, Brasil Abstract—Capacity planning tools aim at monitoring and evaluating computational infrastructures resources in terms of workloads. The Capacity Advisor is a capacity planning tool inside the HP Virtual Server Environment - VSE. This paper describes the development and use of a synthetic workload generator named WGCap (Workload Generator for Capacity Advisor), a flexible synthetic workload generator that generates traces to be imported into the Capacity Advisor for simulating the consumption of resources like CPU demand, memory size, disk data rate and network data rate. The workload generation is reliable and flexible, allowing the generation of workloads based on actual traces or on a significant set of probability distributions and statistical summaries. Index Terms—Synthetic workload generation, capacity plan- ning, random variates. I. I NTRODUCTION Information systems are increasingly present in all activities of our daily lives and in the most diverse areas of knowledge, from health, education and communication to finance, security and entertainment. Given the widespread use of information systems, performance of services provided have been object of attention of service providers, infrastructure managers, application designers and the scientific community. Capacity planning consists on the definition of the necessary resources in a way that the services provided (accomplished) by the systems, meet the appropriate quality levels (response time, level of availability etc.) to the processes that use these services or resources, considering future demands. In fact, for the successful accomplishment of such an activity, the resources should be designed and tuned to minimize the associated costs and to meet the established service level agreement. Capacity planning tools aims at monitoring and evaluating computational infrastructure in terms of workloads (values removed from the tracking of real systems or produced syn- thetically) allowing the analysis of resource utilization and planning of system’s infrastructure. The performance evaluation of systems considering work- loads generated from real traces is an important feature, but a workload generator should also generate workloads based on statistics figures, since the real traces might not be available to the analyst. The reasons why an organization may not make the traces available for analysis encompass confidentiality of content and the non-availability of trustful set of significant traces. Many works related to synthetic workload generation have been conducted over the years in several areas of knowl- edge. [10, 12] have proposed approaches for validating syn- thetic disk request generators. [6] shows the design and use of a synthetic workload generation tool for simulation of web proxy caches, called ProWGen. [14] shows another tool called WebTraff. The tool is an evolution of ProWGen that has an interactive graphical user interface (GUI) that can be used for modeling and analysis of Web proxy workloads. [5] intro- duces SWORD (Scalable WORkloaD generator), a workload generator for testing and benchmarking of high-volume data processing systems. This work concerns the development of a flexible mecha- nism for workload generation applied to a capacity planning tool for virtual environments, the HP Capacity Advisor. The synthetic workload simulates the consumption of main compu- tational resources (CPU demand, memory size, disk data rate and network data rate) and it is generated in a trace format compatible with the Capacity Advisor, so the traces generated can be imported into the tool easily and quickly. This research has two main objectives: development of methodology and computational prototype for supporting workload generation based on actual traces; development of methodology and computational prototype to support the gen- eration of workload based on a significant set of probability distributions and statistics summary. The paper is organized as follows. Section II details the Capacity Advisor (capacity planning tool of the HP Virtual Server Environment - VSE), providing an overview of its operations, architecture and trace format. Section III shows the WGCap workload generation approach proposed, the main scenarios and mechanisms adopted. The section also presents the creation and validation of the synthetic traces. Case studies are shown in Section IV, where the synthetic traces generated by the WGCap are imoported into Capacity Advisor. Finally Section V concludes the paper and presents ongoing and future works. II. CAPACITY ADVISOR This section introduces the environment where the workload generator is applied, where the VSE and the main features of 978-1-4244-2794-9/09/$25.00 c  2009 IEEE SMC 2009