For a long time, BI and data warehouse oa credit tray Source building [sweet potato] Source Forum diguaym.com Contact: QQ: 2152876294 libraries almost always go hand in hand, devotion. If companies want to implement "data-driven decisions - decisions drive the business forward" mechanism, it is necessary to have a data warehouse serves as a central repository for BI query and retrieval, and then analyze and visualize data in BI.
However, data analysis and business decisions to now, companies want to implement data-driven decision-making, whether or not to bypass the data warehouse? Redefine BI and data warehousing in a modern business environment, we can not find a suitable alternative?
Today, we discuss on this proposition, I hope to give you some ideas.
Number of positions: BI behind the engine (or pipe)
Data warehouse: from the literal sense of the data warehouse that is, in order to integrate operational data into a unified environment to provide decision-type data access. Data warehouse concern is to solve the data consistency, credibility, collective ....... these problems, the increasingly complex business data into business operations for the business analyst is easy to use data in the form of ; the ultimate goal of a data warehouse is to make data applications personnel (whether CEO or ordinary analysts) think about how to use the data warehouse data, information and create more value; and not worry about where the data in the data right.
BI (Business Intelligence): BI is to analyze the data and get insights to help a range of methods, techniques and software companies to make decisions. Compared to the data warehouse, BI also includes data mining, data visualization, multidimensional analysis, classification labels and so on. Take for example a multi-dimensional analysis, data warehouse provides the only dimension of the data, but based on some tools, such as Ebay's kylen or IBM's Cognos, etc., can support any combination of user dimensions and metrics within a certain range, that this It rose to the level of decision support rather than "advanced data warehousing" level, that is, the use of the data warehouse, but not the function of the data warehouse.
传统BI项目的构建路径决定了其必须依赖数据仓库才能进行数据分析。比如MicroStrategy，SAP BW，微软 Analysis Server, IBM的Cognos，Oracle的OBIEE，这些传统BI工具不具备使数据集成标准化的能力，数据仓库的存在就是帮助他们建立数据治理结构，解决数据冗余、不一致、错误、无法轻松访问等问题。
另一方面，BI对数据仓库的这种依赖其实存在着极大的缺陷。一般来说，数据仓库通常需要花费高经济成本、时间成本从规划到落地，但创造的价值大多数情况比较有限，ROI较低。搭建成功后，数据仓库也仅支持极少数特定类型的分析，如果企业业务出现调整或者需要处理新类型的数据，届时又将重新面临重大的开发工作。
从现代商业决策视角，重新审视BI与数据仓库的关系
在如今转向服务导向架构(SOA)(*由Gartner提出，以“服务”为基本元素来组建企业IT架构的方式。SOA要解决的主要问题是：快速构建与应用集成，现已成为解决企业业务发展需求与企业IT支持能力之间矛盾的最佳方案。)的技术大背景中，耗费巨大心力进行大规模的数据整合和数据集成操作是否还有必要?构建数仓的收益是否能大于你将付出的成本?
再加上企业数据体量不断提升，业务发展越来越迅速，对快速印证分析决策也提出了更高要求，更多的企业希望能够降低技术设施成本，做到近乎实时地访问操作源数据，在极短的时间内响应用户请求。
于是我们看到了越来越多没有数仓的BI项目。一方面，敏捷BI的兴起，允许用户快速接入各类数据源，无需借助数仓即可实现数据导入-处理-分析的流程。而另一方面，新一代AI+BI智能数据分析平台，则在快速接入、敏捷分析的基础上，实现了更进一步的应用：
- 自带轻量的分布式数据存储与数据流处理模块，提供从数据抽取、数据建模、数据分析，到数据可视化、预警分发的一站式数据分析应用能力;
- 即便不抽取数据，也可实现多数据源的联邦动态分析(联动、钻取、动态参数等交互分析功能)。
In this perspective, to a certain extent can intelligent data under the premise of no data warehouse analysis, however, this is limited to the limited amount of data that small and medium enterprises, does not mean that we recommend directly take data stored on a data analysis platform as data warehouse to use.
Because as the amount of user data business, rising complexity analysis, data analysis platform lightweight data storage and data stream processing module is unbearable enormous pressure calculation, taking into consideration from the perspective of long-term development, there is still need It plans to build a data warehouse or data platform.
Agile enterprises to build strategic analysis of the decision-making framework
Analysis of business prospects for future decision-making structure, depending on the direction of development of business drivers and technology. Today, enterprise data growing exponentially, demand for real-time analysis of intense than at any time in the past, in view of this, how to balance rapid landing and high scalability, the combination of the data warehouse to build enterprise architecture analysis and decision, is still placed in many a huge problem in front of businesses.
In this regard, we recommend the best practices are:
- When the number of positions has not been set up or analysis of ideas not yet formed directly in the BI platform to quickly build analytical applications, rapid feedback, rapid iteration, implement quick win.
- After analyzing the results obtained confirmed service, then the data analysis logic precipitated and gradually solidified to complex data warehouse or data platform inside embodiments, the BI analysis visualization platform only bear pressure and lightweight case data.
We believe that the nature of the data analysis for business development, business decisions and services, rather than creating a bunch of useless visualizations. With this agile mentioned above, quickly confirmed, continuous precipitation process, will ensure that the direction of the enterprise architecture analysis and decision can correct a greater extent on, to get the business end of the recognition, drive business development, creating real business value.