The challenges in building a system of data analytics and big ballsDesigning an integrated foundation is not easy. Extract, transfer, and load (ETL) has been the longest phase of warehouse project data. There are a lot of different solutions, we ETL, sometimes work, sometimes not. If ETL does not perform well, your friends will have random data is not accurate, unreliable and unreliable data, create a system is not reliable and not used.You may think that this can be very easy thing, like a database, but it will become a game, multi version, version updates, release different release cycle is different, different licenses, and different paper allowed based on position. This is some products in a company. It will become more complex, in this case, retail enterprises have tens of thousands of different products.Basic data integration and big ball can be email from unstructured data. They may include semi structural data from the log log. E-mail systems can be distributed in different databases in a lot of data centers in a global project, the integration of more firewalls, or suddenly from the mobile data to another place is a nightmare. In another project log, the system log can be non formatted, formatted, or sold in a mess.There is a reason why big data technology Hadoop encourages mobile data systems such as Apache systems to move to data rather than mobile. Take time to pass a firewall between the mobile data network. You may have lost the data packets to the data file. Degree, believe will become a big problem.The Hadoop NoSQL core and a concept of mobile apps to the data, but this is not easy. If you have 100 different systems, you can add to 100 cases, with an application for each system? Although some people think they can do MDM boss, but in fact there is No. When you have a MDM MDM product, sales, and customer MDM integration or join is not easy, each application system integration or does not mean to join them. It is still a multi barrier system that can't be connected.Even if an enterprise application installs a perfect integration based on a large data, it can be connected to different types of data, and also serious problems can occur. The fact is that you can't suddenly algorithms run in a complex system that the user is using. This may fail, it can make the implementation of the efficiency of slow. It can put all the data there may be security issues. Installing an application requires a large amount of memory space and speed, which can make an old system failure, or even impossible to operate completely in the old system. If it works, it can be in the presence of different systems, or do you choose not to connect the MDM system to the ball?Based on a large data, analytics needs to be innovative, this is the next generation. It uses the technology, in the configuration utility or memory system, such as Apache Cassandra and Hadoop sandbox area, storage system and a new and improved ETL System integration. It is the data structure, non structural and structural sell. Mathematical problems in this difficult part.
đang được dịch, vui lòng đợi..