10 things that big data won't "take the initiative" for the Enterprise

Many companies have high expectations of large data, hope it can solve the long-standing business problems, so that the company more competitive and design, create better products. However, such enthusiasm can easily lead to an overestimation of large data, because the big data "itself" does not bring any value. This article lists 10 large data that will not be "active" for the enterprise, unless the enterprise is more in-depth and detailed analysis and excavation of these data.

1. Solve Business problems

Big data doesn't solve business problems, and businesses still need people to solve them. Only companies that sit down and think about what they want to get from big data before they start using big data can get the business intelligence they're looking for from big data.

2. Help with Data management

IBM claims that the world produces approximately 2.5 quintillion of data every day. Most of them are large data. Predictably, the data being managed within the global enterprise is growing exponentially. Organizations are facing the challenge of managing these data as data accumulates without explicit data retention and usage strategies, especially for large data.

3. Lifting security concerns

For many enterprises, it is still an open topic to determine the safe access to large data. This is because the security practices of large data are not as clearly defined as the system records data. We are in a state where it should work with end users to determine who has access to the level of http://www.aliyun.com/zixun/aggregation/14294.html "> Large data and the corresponding analysis."

4. Solve Key IT skills

Large data database management, server management, software development and business analysis skills are in short supply. This makes it more burdensome for many IT departments that already lack critical IT skills.

5. Reduce the value of legacy systems

Legacy systems are often more valuable than large data, if any. Typically, these legacy systems provide important clues about how to best parse large data and answer important business questions.

6. Simplify Data center

Large data requires parallel processing of compute clusters and a system management that differs from traditional it transactions and data Warehouse system types. This means that the energy, intelligence, software, hardware, and system skills required to run these new systems are also different.

7. Improve data quality

The beauty of traditional trading systems is that these systems are all fixed data field lengths, and comprehensive editing and validation data helps to make the data relatively clean. The big data is different, it's unstructured, and it could be any format. This makes large data quality a big problem. Data quality is critical. Without data quality, you can't trust the results of a data query.

8, verify the existing investment rate of return indicators

The most common way to measure ROI from System records is to monitor the speed of the transaction and infer what this means in terms of income (such as the hotel reservation you receive per minute). Transaction speed is not a good metric for large data processing, and it can take hours or even days to process and analyze large ranges. On the contrary, the best criterion for evaluating the effectiveness of large data processing is utilization, and the results of periodic evaluations should be over 90% (in contrast, the trading system is about 20%). It's important to develop new ROI indicators for big data, because you still need to convince the CFO and other management to prove the value of big data investments.

9, most of the data is useful

95% of the Big data is "noise", that is, no contribution to business intelligence or a small contribution. Filtering out this data for intelligence will be of great use to the business.

10, every time it works

For years, universities and research centers have been experimenting with large data to find elusive answers to studies such as genome Engineering, medical drug research and the identification of the presence of extraterrestrial beings. Although these data analysis algorithms produce some results, more still remain inconclusive. If the uncertainties in universities and research environments are tolerable, then that is not the case in the corporate environment. This is what it and other key decision makers need to anticipate.