Page 50 - Kaleidoscope Academic Conference Proceedings 2024

2024 ITU Kaleidoscope Academic Conference

resource configuration model for the inspection rule. b) The orange represents unresolved issues, and blue represents the
rule can be translated into the semantics expressed by total number of inspection rules.
Formula 1. c) Includes information about the fields involved
in the inspection rule. d) Contains a clear inspection
execution expression for the expression engine to execute. e)
The expression includes engine extension semantics to
support specific business rules. f) For each inspection rule, it
includes handling suggestions to facilitate closing the loop
on identified issues.

(3) The presentation of inspection results. In the end, we
need to present the inspection results to users and provide
result export and snapshot functionality. When designing
persistent storage, we need to meet the following conditions:
a) Data snapshots to be inspected at the inspection time, for Figure 5 – Tendency chart in CMDB inspection system
self-certification of results. b) Inspection model, start time,
end time, and other basic information. c) Inspection results. We also provide filtering options for different resource pools
in the 'IT Cloud Regional Center' and 'IT Cloud Provincial
3. RESULTS AND DISCUSSION Nodes.' Additionally, there are multi-time dimension query
options for the past month/half-year/year, and a
3.1 CMDB scenario illustrations customizable time range. As shown in Figure 6. In the
accuracy rate table, we calculate the accuracy of data items
At present, CMDB in the product environment has a total of reported to CMDB at the granularity of natural months for
27 models with a total of 779 inspection rules. With the different resource pools. This provides a basis for evaluating
increase in business requirements, the number of inspection the scores given to reporting sources by the IT company,
rules is also continuously rising. We can showcase the details further promoting the improvement of data quality through
of problematic data after inspection for different models the CMDB inspection system.
(bare metal, hosts, virtual servers, etc.), provide multi-
dimensional filtering functions, and offer result export
capabilities. The specific interface and functionalities are
illustrated in Figure 4.

Figure 6 – Accuracy statistics for resource pool
reporting in CMDB inspection system

3.2 Accuracy

Figure 4 – CMDB inspection results display interface Through the analysis of product data, this method has proven
effective for 49 different reporting sources, resulting in
The diagram illustrates the results of an inspection for the varying degrees of accuracy improvement. The accuracy is
'Bare Metal' model, conducted on 2023/03/30 4:14:37. From consistently maintained above 90%, with an average
the inspection results, a total of 3602 issues were identified. improvement of approximately 12%. This has effectively
Each row displays the details of problematic data, including enhanced the accuracy of CMDB data reporting, calculated
statistical information such as the occurrence time, duration by the formula Accuracy = (1 - (cumulative number of
in days, and the inspection rule triggered by the data. The problem data entries in the current month + unresolved
inspection results support export functionality, facilitating problem data entries from the previous month) / end-of-
the communication of issues to relevant personnel and month server scale) * 100%. Figure 7 illustrates the
driving issue resolution. Regarding the progress of issue significant improvement in data quality for 10 resource pools.
resolution, we offer features such as trend charts and For instance, the Jilin resource pool data source, from
accuracy. As shown in Figure 5, the trend chart displays the October 2022 to February 2023, saw an accuracy
quantity over time, with the x-axis representing dates and the improvement from 32% to 99% after the inspection.
y-axis representing the quantity. Green indicates new issues,

– 6 –