The Architecture And Design Of ODS/EDW System For Large Telecom Operators

Along with the gradual maturity of all sorts of standards and technologies in IT field, software packages that support various business processes are constantly emerging. For the increasing competition and the dwindling profits in telecommunications services industry, major telecom operators start in full swing for the construction of enterprise information system. The construction of operational data store (ODS) and enterprise data warehouse (EDW) is not only the core part of the enterprise data architecture (EDA) in large telecom enterprises, but also the fundamental assurance of the enterprise data quality.This paper will focus mainly on the key and difficult problems of the construction of ODS and EDW in some large telecom operator, including the establishment of the overall system and detailed framework, client model, index system, various clearing, sharing and resolution rules model, enterprise data model, norms and needs of data extraction, data conversion and data loading, perfect data audit system, facilitated data verification methods and so on.The establishment of the overall ODS/EDW system and detailed framework is very difficult. Although there are always rough patterns as references for the overall structure, there are still lots of things need to be considered about the enterprise actual fact. Many architectures fail because of their failure in balancing all kinds of complex business logic of actual enterprises. This paper takes into account various factors and gives the rational ODS/EDW framework.With considerations of history and reality, the establishment of client model and the index system should also meet with the requirements of modern data warehouse. The paper gives a practically feasible and properly abstracted and simplified client model and index system according to the analysis of existing conditions. Various clearing, sharing and resolution rules are the most complex logic in ODS/EDW system, and they have a high demand of accuracy and verifiability. In the whole system architecture, they belong to the income compilation model. This paper analyzes and abstracts various pretreatment, SP settlements, internal assessments and other rules, then finally gives a reasonable design.Enterprise data model needs to be able to abstract and adapt to the whole enterprise’s business entities and processes. It should be universal and also on behalf of the characteristics of enterprises. This paper analyzes seven data theme domains and eight business concepts in enterprise data model, and introduces the important data entities. Data extraction and conversion, data loading are the main approaches to acquire the source data of ODS/EDW system. These steps must follow certain methods and norms, which ensure the orderly acquisition and quality assurance of the ODS/EDW system data.Data audit system and data verification methods are issues that must be considered at the beginning of the construction of ODS/EDW system. This paper creatively proposes an entire audit system and summary the significance of audit bus, and builds the verifiable mechanism in the ODS/EDW platform. This fundamentally guarantees the data quality of the entire ODS/EDW system, and provides convenient methods and channels for various business units.