In the last article ( Data Center Construction Series : What is a Data Center), we talked about what a data center is, and we also know the huge value of a data center. Can we start building a data center? I think, before officially entering the construction of the data center, let's talk about what kind of enterprise is suitable for building the data center, so that everyone can rationally analyze and choose according to the actual situation of the enterprise, so as to prevent blindly following the trend from causing huge losses. 1. Building a data center Common data pain points of former enterprises Due to work reasons, they participated in multiple data middle-office projects. During this process, I found that many companies usually have a series of pain points before building data middle-offices. To sum up, they can be summarized as the following 5 major kind.
The caliber of indicators is not uniform In mobile number list the two reports, the indicators with the same name [Sales] display different results. The business suspects that there is a problem with the data, so they find a development investigation. The inspection results show that one of these two indicators includes tax and the other does not. When business personnel are faced with these indicators, if they do not know the business caliber of the indicators, it is difficult to use these data. 2. Long demand response time As demand continues to grow, operations and analysts complain that the delivery time of requirements is too long to meet the agile R&D requirements for data in a rapidly evolving and changing business.
Inefficient fetching With the continuous growth of data, in the face of massive data tables, it becomes more and more difficult for operators and analysts to accurately find and understand data. A large number of temporary data retrieval work can only be completed by data research and development, which makes data research and development impossible to focus on. In the construction of the data warehouse model, a vicious circle of [incomplete data—R&D is busy with various temporary data retrieval requirements—incomplete data] is formed. 4. Poor data quality From time to time, data results are calculated incorrectly, leading to wrong business decisions. Data bugs occur frequently, and fault tracing and recovery often consume a lot of time.