資料管理

Data management is a collection of concepts and practices used to manage data produced in organizations. Data management has grown more complex while the notion that data as a valuable resource to be refined has grown in popularity. In many cases, enterprise business operations without data analytics are simply not a feasible strategy to remain competitive.

A general view of data management shows it to have two layers: a technical layer, where the actual processing of raw data occurs by computer processes associated with data collection, transformation, and analytics, many of which can be automated; and a non-technical layer, where non-technical personnel are engaging more and more with data to draw insights that assist in their everyday tasks.

Data management as a discipline has many sub-categories, and in many cases requires their own specialization.

These subcategories address the forces and factors that influence an organization’s data management strategy:

• Data Governance — Data governance addresses the political nature of data.
  
• Data Architecture  — Data architecture addresses which data is collected and stored.
  
• Data Modeling and Design — Data modeling and design addresses the technical structuring of a data model and designing databases that fit data to that model.
  
• Database and Storage Management — Database and data storage management address the active management, maintenance, and assurance of database infrastructure.
  
• Data Security — Data security addresses guarding data assets against bad actors who can steal, change, and destroy data.
  
• Reference and Master Data — Reference and master data addresses the consistency and continuity of disparate data sources across the organization.
  
• Data Integration and Interoperability — Data integration and interoperability addresses the technical movement and transformation of data from multiple sources in a unified view.
  
• Document and Content Management — Document and content management addresses the cataloging and managing of information held in various business documents.
  
• Data Warehousing and Business Intelligence — Data warehousing and business intelligence addresses data mining, data analytics, and data warehousing.
  
• Metadata — Metadata addresses the discovery and management of data about other larger data sets.
  
• Data Quality — Data quality addresses the distillation of raw data into usable forms.

There are many facets of data management, as defined by its subcategories, however, the overarching benefit is to increase control and visibility into data assets within an organization. Without any form of data management, companies flounder in operations and growth, falling behind in competitiveness, and eventually out of the market. Modern data management systems, though, are ready made for many organizational cases, and provide several benefits that branch from improved data control.

• Improved Data Quality and Accuracy — Impurity is a characteristic of most natural resources, one that data emulates. Using effective data management helps to clean and distill raw data and improve quality, ensuring that analysis can begin with the most accurate and usable data.
  
• Reduced Time and Cost — Companies burn resources managing their data, and burn time and money trying to overcome the blindspots that quality data insights may have foreseen. Modern data management software reclaims time and costs in collecting, managing, protecting  and analyzing data. The insights drawn from data analysis bolster other processes and systems, improving performance and efficiency, and improving the effectiveness of time and dollars spent.
  
• Eliminated Data Redundancy — Redundant data introduces risks and weaknesses into an organization’s data system, typically in the form of data inconsistencies. Data inconsistency occurs during data operations, like the insertion, deletion, or adjustments to data entries, that result in anomalies that degrade data integrity.
  
• Guaranteed Data Compliance — Regulation of personally identifiable information has made data compliance a top priority for many businesses. Failure to comply can be met with penalties as well as reputational damage to the brand. Data management equipped with data governance functionality can help track and guarantee data compliance and protect the business.
  
• Informed Business Decisions — A key design principle in data management software is the unification of disparate data and information for analytics. By bringing the management of data and data sources under central monitoring and control, organizations are provided a holistic view up-to-date views of their data usable in real-time operations.
  
• Single Source of Truth — All the features of data management come together to provide a single source of truth (SSOT) for the company. SSOT architecture is structured in such a way that every data point is managed and controlled from a central point, and data then can become normalized, or made canonical. Changes made to this canonical version of data, then ripple throughout the entire data system, providing the latest version to every connected business system.

Cloud data management combines the advantages of cloud services with the power to manage data across cloud platforms. Cloud advantages include resource scaling, disaster recovery, anytime anywhere access, backup and long-term storage, and cost controls. Cost controls are particularly beneficial for business, either small or enterprise, and grant both of them the ability to pay for resources as needed. Typically cloud providers will be responsible for maintenance in the cloud relieving those worries from businesses.

In other instances, organizations can integrate their own resources.

Multi-environment compatibility is another important capability of most cloud data management vendors. Data can be shared and integrated across private and public clouds, providing access to on-premise storage.

Data management requires a thorough look at the data requirements of a specific domain. However, general best practices, like the ones below, can help circumvent potential challenges.

• Establish a data discovery layer — A data discovery layer enables the searchability of all data sets.
  
• Streamline data science tasks — Data transformation overhead should be streamlined as much as possible to ensure real-time data analysis.
  
• ·Automate and use AI — Automation and AI can help teams proactively monitor database queries and optimize indexes. Automation emancipates team schedules and enhances efficiencies.
  
• Leverage data discovery to maintain compliance — Data discovery enables teams to maintain up-to-date compliance with multiple jurisdictions. As regulations increase globally, this auditing task will become more vital.
  
• Use converged databases — Converged databases allow native support of multiple disparate data types and platforms.
  
• Align your database platform with business needs — The main goal of analyzing data together is to see a larger picture of the business. Speed, accuracy, and reach need to match those of the business. Enterprises may need split second analysis, while smaller businesses not as much.
• Use a common query layer — Like converged databases, using a common query layer enables data scientists, and others, to access data from any source without needing to know its location.