Amazon is giving users a comfort blanket to quickly pull large amounts of data out of its cloud.
AWS Snowball is a physical appliance, with which customers can store and ship large sets of data from their private data centers to the provider's public cloud. The nascent service was previously limited to importing data, but now, users can have the box mailed back to them to quickly export data sets from the cloud.
Amazon executives emphasized Snowball when it launched at the annual AWS re:Invent conference last October. The company envisions a two-year window to get customers to move their data to AWS, and the turnkey Import/Export Snowball service eases that transition for customers, said Robert Mahowald, vice president at IDC.
"It's clear that they're trying to get customers on their sticky cloud, so Snowball is a huge part of that," Mahowald said.
It's unclear how much traction Snowball has achieved, but customers are enthused, noted Paul Miller, senior analyst at Forrester Research.
"I do get the sense that an awful lot of customers are using it to push data into Amazon," he said. "The timing required and the network costs required to move large volumes in has always been a [hurdle] in that move to the cloud."
The ability to import and export from AWS with physical devices dates back to 2009, but Snowball streamlines the service by providing a plug-in appliance mailed directly to customers, with the capacity to store up to 50 TB of data. The data is encrypted and sent to data centers in either Amazon's U.S. East or U.S. West regions to be transferred to customer-specified Simple Storage Service (S3) buckets.
The new export feature allows customers to take those same steps in reverse, with the only notable difference being a charge of $0.03 per gigabyte. There is no transfer fee for inbound data. The service fee per job, whether imported or exported, is $200.
Snowball practicality goes beyond emotional support
Costs aside, the benefits of making it a two-way street could be as much psychological as they are practical, according to Miller.
For customers, downloading their data over a network can be timely and expensive, so the export feature can be an insurance policy -- and, arguably, an important comforting message from Amazon that its users can get their data out of the cloud, if needed, he said.
As a more practical application, some customers could benefit from the export feature to process jobs over short periods of time and pull them back on premises for uses, such as sequencing data, he added.
Jobvite Inc., a San Mateo, Calif., recruiting software company, is an AWS customer that doesn't ship a lot of on-premises data to Amazon's cloud, but there is tremendous value for companies that need to transfer hundreds of terabytes or petabytes of data, said Theo Kim, senior director of software as a service operations.
"It's faster than doing data transfer over the Internet, and also doesn't rack up the Internet charges," Kim said. "It's definitely a great system ... they'll ship you a device, you can copy the data on to the device and ship it back with a pretty good [service-level agreement]."
That said, making the Snowball service bidirectional will only benefit users who want to get petabytes of data in or out of the cloud, Kim said. And even though S3 is durable, there are scenarios, particularly for database or MapReduce data, where a user may want to export the data to back up somewhere outside of AWS.
Trevor Jones is a news writer with TechTarget's data center and virtualization media group. Contact him at email@example.com.
New AWS migration tools could lead to lock-in
Consider these five AWS data migration methods
Migrate from VMware to AWS with these tools