Enterprises have a variety of choices when deploying big data applications in the cloud, and the amount of choices...
By submitting your personal information, you agree that TechTarget and its partners may contact you regarding relevant content, products and special offers.
continues to rise. Having more choices means IT teams have some important decisions as well. Should they rely on a cloud appliance or build the necessary infrastructure themselves?
The first wave of big data cloud appliances started to hit the market last year. Cloudera Inc., EMC Corp., Yahoo spinoff Hortonworks Inc., Infobright Inc., MapR Technologies and YottaStor all entered the game.
Using a big data appliance could have several advantages for enterprises, first of which is convenience. But often the high cost of third-party big data tools refutes that benefit.
"Vendors configuring these types of appliances have deep, direct experience with big data analytics, so they try to put together appliances with the appropriate storage, memory, bandwidth -- the networking aspects required of big data are often overlooked -- and software," said Evan Quinn, analyst at Enterprise Strategy Group. As a result, IT shops no longer have to determine how to configure the system.
In addition, appliances often come wrapped with management software and maintenance services that not only focus on appliance hardware but also help enterprises meet business objectives.
"To be successful, a big data project needs to deliver a compelling business benefit," said Charles Zedlewski, vice president of products at Cloudera. "Sometimes, enterprises focus on their technical elegance and do not have clear business objectives."
To be successful, a big data project needs to deliver a compelling business benefit.
Charles Zedlewski, vice president of products at Cloudera
There are some potential downsides to big data tools, including security concerns. Companies involved in financial services and health care may not want to put their data on a third-party provider's cloud.
Additionally, performance can suffer. Big data vendors typically offer generic platforms that may not be optimal for a company's unique application.
Cost can be another issue. "There are huge variations in price, starting at $10,000 and sometimes [reaching] well over $1 million," said Zedlewski.
While big data cloud appliances simplify deployments, they are not simple drop-in systems. There are several support and maintenance issues. IT shops need to track versioning issues; they still have to get data into a big data tool and mine results from the appliance. Therefore, enterprises need to put procedures in place to back up and migrate data as it changes.
ESG found that, while big data cloud appliances represent a minority of implementations -- about 10% to 15% -- that number is rising. "Over the coming 12 to 24 months, I expect big data vendors to come up with more compelling cloud appliance offerings," added Quinn.
Paul Korzeniowski is a freelance writer who specializes in cloud computing issues. He is based in Sudbury, Mass., and can be reached at firstname.lastname@example.org.