No. 1: You will need to think about big data
Big data analysis got its start from the large Web service providers such as Google, Yahoo, and Twitter, which all needed to make the most of their user-generated data. But enterprises will big data analysis to stay competitive and relevant.
[ Get sage advice on IT careers and management from Bob Lewis in InfoWorld's Advice Line blog and newsletter. | Get expert advice about planning and implementing your BYOD strategy with InfoWorld's 29-page "Mobile and BYOD Deep Dive" PDF special report. ]
You could be a really small company and have a lot of data. A small hedge fund may have terabytes of data, said Jo Maitland, GigaOm research director for big data. In the next couple of years, a wide number of industries -- including health care, public sector, retail, and manufacturing -- will all financially benefit by analyzing more of their data, consulting firm McKinsey and Company anticipated in a recent report.
There is an air of inevitability with Hadoop and big data implementations, said Eric Baldeschwieler, chief technology officer of Hortonworks, a Yahoo spinoff company that offers a Hadoop distribution. It's applicable to a huge variety of customers. Collecting and analyzing transactional data will give organizations more insight into their customers' preferences. It can be used to better inform the creation of new products and services, and allow organizations to remedy emerging problems more quickly.
No. 2: Useful data can come from anywhere (and everywhere)
You may not think you have petabytes of data worth analyzing, but you will, if you don't already. Big data is collected data that used to be "dropped on the floor," Baldeschwieler said.
Big data could be your server's log files, for instance. A server keeps track of everyone who checks into a site, and what pages they visit when they are there. Tracking this data can offer insights into what your customers are looking for. While log data analysis is nothing new, it can be done down to dizzying new levels of granularity.
Another source of data will be sensor data. For years now, analysts have been speaking of the Internet of Things, in which cheap sensors are connected to Internet, offering continual streams of data about their usage. They could come from cars or bridges or soda machines. "The real value around the devices is their ability to capture the data, analyze that information, and drive business efficiencies," said Microsoft Windows Embedded General Manager Kevin Dallas.
No. 3: You will need new expertise for big data
When setting up a big data analysis system, your biggest hurdle will be finding the right talent who knows how to work the tools to analyze the data, according to Forrester Research analyst James Kobielus.
Big data relies on solid data modeling. Organizations will have to focus on data science, Kobielus said. They have to hire statistical modelers, text mining professionals, people who specialize in sentiment analysis. This may not be the same skill set that today's analysts versed in business intelligence tools may readily know.
Such people may be in short supply. By 2018, the United States alone could face a shortage of 140,000 to 190,000 people with deep analytical skills, as well as 1.5 million managers and analysts with the know-how to use the analysis of big data to make effective decisions, McKinsey and Company estimated.
Another skill you will need to have on hand is the ability to wrangle the large amounts of hardware needed to store and parse the data. Managing 100 servers is a fundamentally different problem than handling 10 servers, Maitland pointed out. You may need to hire a few supercomputer administrators from the local university or research lab.
No. 4: Big data doesn't require organization beforehand
CIOs who are used to rigorously planning out every sort of data that would go into an Enterprise Data Warehouse (EDW) can breathe a little easier with big data setups. Here, the rule is collect the data first, and then worry about how you will use it later.
With a data warehouse, you have to lay out the data schema before you can start laying in the data itself. "This basically means you have to know what you are looking for beforehand," said Jack Norris, vice president of marketing for MapR. As a result, "you are flattening the data and losing some of the granularity," he said. "Later on, if you change your mind, or want to do a historical analysis, you've limited yourself."
"You can use a [big data repository] as a dumping ground, and run the analysis on top of it, and discover the relationships later," Norris said. Many organizations may not know what they are looking for until after they've culled the data, so this kind of freedom "is kind of big deal," he said.
No. 5: Big data is not only about Hadoop
When people talk about big data, most times they are referring to the Hadoop data analysis platform. "Hadoop is a hot-button initiative, with budgets and people being assigned to it" in many organizations, Kobielus pointed out. Ultimately, however, you may go with other software.
Recently legal research giant LexusNexus, no slouch at big data analysis itself, open-sourced its own platform for analysis, HPCC Systems. MarkLogic has also outfitted its own database for unstructured data, the MarkLogic Server, for Big Data style jobs as well. Another tool gaining favor is the Splunk search engine, which can be used to search and analysis data generated by machines, such as the log files from a server. "Whatever data you can extract from your logs, there is a good chance that Splunk can help," noted Curt Monash of Monash Research.