Takashi Yanase

Assessment method of operational procedure for runbook automation

The large demand for cost reduction of system operation has made the shift to runbook automation essential, because runbook enables both the operational cost to be reduced and operational quality to be improved. Because of the reduction in the cost of automation development, it is important to assess the operation manuals in advance and to select the operational tasks to automate. In the assessment, the development volume of automatic operational procedure should ideally be estimated so that it corresponds to the contents of all operation manuals. However, there was a problem that the estimation was rough in the current assessment carried out manually by sampling research. Thus, we propose an assessment method that can estimate more accurately from all operation manuals by considering how the common procedures are integrated in the actual automation development. The proposed method was shown to be practical by comparing the estimated value of the development volume of an automatic operational procedure, calculated by applying the method to actual operation manuals, with the result value. Moreover, the proposed method enables to forecast beforehand whether operation tasks can be automated easily by calculating the payback period on the basis of the estimated development volume of automatic operational procedure.

Study on efficient analyzing method of operation manuals for runbook automation

Recently, server consolidation using virtualization is increasing and needs for IT system operation automation are increasing, as well. To shift to automated operation, it is necessary to analyze operation manuals and to standardize procedures by selecting suitable ones. In an environment of server consolidation using virtualization, operation manuals for different systems are gathered for analysis. However, analysis of operation manuals for automation requires high computational complexity. Therefore, analyzing them manually is almost impossible and it is one of the causes that prevent migration from manual to automated operation. In this paper, we propose efficient and comprehensive analysis focused on the characteristic wording of operation manuals and apply natural language processing and data processing algorithms. We apply this approach to a real case of automated operation migration and show that it can accurately and quickly process the entire operation manual.