Every business and organization must consider its own needs and resources when figuring out which issues are most important for big data implementation. However, there are a number of principles that are generally considered critical for this kind of adoption of technology.
|Webinar: Big Iron, Meet Big Data: Liberating Mainframe Data with Hadoop & Spark
One of the biggest questions is implementation and the amount of disruption it will cause. Users of big data systems always have to compare what they are about to use to what they are currently using. In many cases, disruption is the deciding factor in whether big data resources are going to boost productivity and profits, or send a business crashing down due to insurmountable hurdles with implementation. Vendor support (or lack of it) has a lot to do with this, but businesses also have to look at the learning curve for technologies, how much they would change the operations of legacy systems, and in general, whether the changes are something that the enterprise can handle.
Another major question is which data is most valuable to a business or organization. By examining the value of different data sets, those intending to implement big data can set the scope of their project. Without these kinds of guidelines, big data projects can get bloated and overwhelmed in an enterprise. Experts recommend focusing on the specific data sets that will give the most value, without getting bogged down in casting a wider net.
A corollary issue here is the use of structured and unstructured data. Business leaders can look at the levels of difficulty of getting different bits of data into a big data context like a data center. For example, already formatted data sets can be easily digested, but some other pieces of data may need extensive manipulation to get them into a useful format, and it may not be worth it.
Adopters will also have to look at advanced handling for big data. Big data systems are defined as those that are difficult to handle with basic and simple hardware and software infrastructures. That means the adopters need to have adequate talent and resources on hand to find ways to use the big data sets that won't cause network congestion or otherwise create bottlenecks in operations.