【 Weak current College 】 implementing data virtualization should avoid common mistakes of ten

Data virtualization to overcome the complexity of the hardware and software capabilities to enterprises to increase IT flexibility and significant cost savings provided an excellent opportunity. As more companies seek such benefits, data virtualization is fast becoming the new ideas from mainstream applications. Here are the early users often make ten mistakes, hope as some objectivity of lessons to help enterprises accelerate data virtualization may bring potential benefits.

Error 1: too many virtualization

Data virtualization and storage, server and application virtualization, providing an excellent profit gains. For example, an energy company uses data virtualization to real-time oil field data and every night of the integrated warehousing information together took daily oil production increases several thousands of barrels. A financial firm to the new application development time is reduced by 50%. At the same time, another financial services company annual savings of $ 2 million in business intelligence and reporting costs.

However, data virtualization not every data integration solution. For example, when a consumer application needs from multiple angles for analysis, or when the data in the consumer need for major change before, physical data integration is a better way.

In order to avoid any specific development projects too much for virtualization, first of all to better understand the business, data resources and data characteristics of consumers.

Error 2: Virtualization is not enough

The opposite of the first error is virtualization. According to the routine, instead of seeking the best way is for everyone to do. In the 1990s, the development of physical data integration in a single, consolidated storage and special ETL (extraction, transformation and loading) middleware software. After the year 2000, ETL has become the default sample data evidence integration. However, it should be the only method you @

Virtualization will increase a lot of low costs, since the physical data integration needs longer solutions, more expensive development and operational costs and lower business and IT flexibility (as containing an additional cost). Fortunately, avoid the wrong approach is the integration of the data in the decision-making process requires careful analysis and definition, to ensure the best solutions to meet these requirements, instead of the traditional approach to promote this decision.

Error 3: error mixed opportunity

In many cases, the best data integration solution is virtual and physical methods. There is no reason to lock this way or that way. Figure 2 shows examples of mixed use, and on these examples.

@ Physical data warehouse/data mart programme extended or: this is the extension of existing programmes, such as the current operational data to historical library.

@ Physical storage, Mart and/or combined storage: this is the multiple physical consolidation resources combined together, such as a merger of two or three sales data set.

@ Data warehousing and/or data marts prototype products: this method is the introduction of a new warehouse or data mart prototype products, accelerate early phase of implementation, thereby into larger business intelligence initiatives,

@ Data warehousing and/or data marts source data access: this is the data warehouse and Data Mart to provide virtual access to the data source of a method, such as current ETL tool is not easy to provide technical support for XML or packaging applications.

@ Eliminate data marts: this is the thing with the virtual elimination or replacement of physical data marts for a method, such as by providing convenient and cost-effective virtualization choice block malicious data mart.

Error 4: Assuming perfect data is a prerequisite for

Poor data quality is the enterprise of a widespread problem. Although the correction and improvement of the source data is the ultimate goal, however, we in the physical data consolidation consolidation phase of conversion and clean warehouse data will still be left some data is not processed.

When data quality issue is the reaction of various system implementation details of the simple format conflict problems, data virtualization solution can easily solve these common data conflict, will not have any images on performance. Some examples include, in a source system Part_id field read as VARCHAR, but the same field in another source system read as INTEGER. Or in a system of sales area and another system that does not match the sales area. If you need to do the heavy lifting of cleanup work that runs with professional data quality solution is usually to meet business requirements while open data virtualization.

Error 5: it is expected that adversely affect your operating system [/V]

Although the operating system usually in virtualization data using one of the main data source, the system's run-time performance is generally not affected by the results. Therefore, designers have the physical storage of the data volume and every night for the extraction, transformation, and loading operations data throughput. When you use virtualization approach, designers should consider the final solution in every actual query when the query is the amount of data and the frequency of query execution. If these queries are relatively small (for example, 1 million rows) and the broad scope of the query (involving multiple systems or table), or run times are not so frequently (such as running hundreds of times a day), then the operating system image is lighter.

System designers and architects are expected to produce a negative image of the operating system is generally underestimated the latest virtualization solutions. Use some time calculation required data load helps avoid potential for operating system images in the wrong judgment.
Error 6: not simplify problem

When enterprise data environment is very complex, develop complex data virtualization solution is usuallyUnnecessary. Most successful data virtualization project are divided into smaller components, each component part of the solution to the entire demand. This simplification can use two methods: using the tools and the use of appropriate integration components.

Data virtualization tools to help troubleshoot data consolidation of three basic challenges:

1. data position: data at multiple locations and sources.

2. data structure: data are not always in the format required.

3. data integrity: data often need and other information together makes sense.

Data virtualization middleware software can solve the problem of data location, method is to have all of the data seem to come from the same place, instead of the actual storage of these data.

Data extraction can simplify the complexity of the data, the method is to convert the structure and syntax of data into a business solutions developer easier to understand consumer and business solutions can be reused and Web services.

Data combined with the data-binding together to form a more meaningful business information, a customer of a single view or a combination of inventory balance.

Success of the correct size of the data integration component needs smart decomposition of various requirements. Recently, including CompositeSoftware company, five manufacturers of experts issued a "SOA implementers Guide" offers three levels of virtualization data services, allowing designers and architects design are smaller and easier to manage data integration components:

@ Physical services: physical service is located in the data source, the data is converted to a higher level of service to easily consume format.

@ Business services: business services reflects the conversion logic of most of the functionality, the data from the physical format into the format required by the business.

@ Application services: application services use commercial services to the consumer application optimized for data. In this way, solutions developers to follow these simple and focused on the design of data services to simplify the current development efforts to provide greater reuse and flexibility.

Error 7: SQL/relational databases and XML/hierarchical data as isolation of shaft

Historically, data integration has been focused on supporting business intelligence application requirements, and process integration focus is optimizing business processes. These two different methods to produce a different schema, tools, middleware software, methodology, teams, etc. However, today's data virtualization middleware software typically suitable for relational and hierarchical data, the key to isolate the data format is wrong. This requires a combination of SQL and XML is very important when, for example, when the external payroll processor data and internal sales force automation system for relational data together in a sales representative's performance Portal provides a single window of XML data.

Whether your data type, a unified approach not only to provide a better solution, and that developers and designers also have access to their traditional core technology areas of experience.

Error 8: using the wrong infrastructure implementation data virtualization

In SOA (service-oriented architecture) environment data services loosely coupled data virtualization is an excellent application. Therefore, data virtualization, SOA is the most frequently used case. However, sometimes for when to deploy enterprise service bus middleware software and when to use information server design and run data services also exist in some confusion.

Enterprise service bus is good at coordinating the various transactions and data services. However, enterprise service bus does not support the onerous task of data features such as high-performance query, complex joint, XML/SQL conversion, the current enterprise application software to use the functionality in the instance. On the other hand. Data virtualization tools provide a convenient, efficient data services development environment and high-performance, high reliability runtime information server to meet design and run-time requirements. Then, an enterprise service bus according to requirements of these services.

Error 9: data virtualization personnel and process isolation

Along with physical data integration techniques and methods of mature to integration Centre (ICC) as well as best practices and processes, technical support organisations has already increased support. These centres to improve developer productivity and optimization tools to use, reduce project risk, and do other things.

That cannot or should not use these ICC support data virtualization is wrong. Through the use of data virtualization, ICC in personnel and process resources with the ability to put data virtualization technology combining value.

Error 10: does development and sharing of benefits

Although data virtualization can speed up the development of new, perform faster iterative changes, reduce development and operational costs, but that these benefits can own self-promotion is wrong, especially in new technology investment requires a rigorous review of difficult business period.

Fortunately these benefits can (and should) measuring and shared. The following is the complete this task some ideas:

@ First to use previously introduced virtual and physical integration tools identify some data virtualization candidates for the project as a pilot project.

@ Design these projects Department and the development phase, tracking it uses data virtualization, and it uses the traditional physical methods all the time.

@ Use this save time of calculating the value of two additional points: solutions reduce time and savings in development costs.

@ To measure the value of the life cycle, the forecast because of virtualization additional physical data storage savingsOperational costs.

@ For faster recovery from failed repair normal and enhanced development activities are expected to save the development life cycle costs increase the hardware cost of operation.

@ Last these results of the pilot project on integration of the future project results deduced, and put such inference to business and it leaders.

Industry analysts agree that the best practices of leaders from contains virtual and physical data integration tool portfolio lessons for today's enterprises changing information needs. A wide range of industries and Government agencies in more than one application instance illustrates the benefits of data virtualization. These benefits include reduced solution launched by time, reduce implementation and maintenance of the entire costs, and adapt to change and greater flexibility. Through the familiar with the need to avoid the common mistakes that businesses will be in data integration infrastructure in successfully implementing data virtualization requires wisdom to harvest data virtualization benefits.