Availability modeling and validation methodology for RS/6000 systems

Author

Ahrens, George ; Chandra, Arun

Author_Institution

IBM Corp., Austin, TX, USA

fYear

1999

fDate

18 -21 Jan 1999

Firstpage

305

Lastpage

309

Abstract

In this paper we show the availability modeling and validation methodology for RS/6000 systems. The availability modeling methodology is used both for prediction of availability characteristics of systems under development, and for the prediction of availability in the special bids process. The core of our methodology is the availability modeling tool-System Availability Estimator (SAVE)-currently being used to predict availability measures. Our availability validation methodology has both a conceptual validation and an empirical validation component. Conceptual model validation is done via model walkthroughs and by talking to the designers. The empirical model validation methodology is performed using tracking data. Concluding, our availability modeling and validation methodology allows us to develop and configure higher availability systems which are critical in the server computer market space

Keywords

fault tolerant computing; reduced instruction set computing; software packages; RISC System 6000; RS/6000 systems; SAVE; System Availability Estimator; availability characteristics prediction; availability measures prediction; availability modeling; availability modeling tool; availability validation; model walkthroughs; server computer market space; tracking data; Application software; Automation; Availability; Current measurement; Data mining; Feeds; Maintenance; Predictive models; Steady-state; Workstations;

fLanguage

English

Publisher

ieee

Conference_Titel

Reliability and Maintainability Symposium, 1999. Proceedings. Annual

Conference_Location

Washington, DC

Print_ISBN

0-7803-5143-6

Type

conf

DOI

10.1109/RAMS.1999.744136

Filename

744136