The AWS Import/Export Snowball appliance enables enterprise IT teams to move large amounts of data to or from the AWS cloud. Using the device can help companies avoid lengthy data transfer times, security risks and network charges incurred with moving data. However, it also has some limitations.
AWS' petabyte-scale data transport device can move 50 terabytes (TB) of data at a time and offers internal and external security, including tamper-resistant enclosures and 256-bit encryption.
It's an effective way to support one-time data imports or workflows for batch processing, said Steve Pao, CMO at Igneous Systems, a Seattle-based company that extends on-premises data centers into the cloud.
To move large amounts of data to the cloud, companies have three primary options: transfer across the wire, buy multiple external hard drives to store data and then ship them to AWS, or use Snowball.
Even with high-speed internet connections, transferring large amounts of data to or from AWS can be slow. Companies order the Snowball appliance, plug it into their network and then ship it back to AWS. The Snowball appliance provides an all-at-once transfer process that mitigates the need for cross-net monitoring -- monitoring your load across the internet. With Snowball, companies don't have to ensure that every packet made it to AWS.
Snowball is not a cure-all
AWS Snowball hasn't solved every data transport problem in the cloud.
Companies often struggle with having a data-transfer plan prior to ordering Snowball. Because of this, companies can incur additional and often substantial costs as the appliance remains on-site -- and unused -- for more than 10 days.
To mitigate this, have a plan on what data needs to be transferred to Snowball and also who is going to perform the transfer -- before Snowball arrives. Enterprises should also have an agreed upon maintenance window to get the work done.
The cost of the Snowball appliance can also be prohibitive. With $250 per 80 TB of data and only a day or two turnaround time, enterprises need to consider transfer time and personnel costs to evaluate Snowball. But the scales tip toward Snowball starting at 2 TB or 3 TB.
Snowball isn't a one-size-fits-all approach. For example, Snowball Edge, a variant on the Snowball concept, is a 100 TB data transfer device with onboard storage and compute power.
"With its built-in Lambda support, it is a step in the right direction to do some inline processing of data," Pao said.
But the device cannot process data while in transit. Still, Snowball Edge has all the same benefits in its favor as the original Snowball, and it adds some features that increase its value.
But while Snowball is secure, companies cannot remove the local data until it is verified on Amazon Simple Storage Service, which may not be acceptable to companies that want to remove data immediately.
The new clustering abilities of Snowball Edge may give users more confidence that their data is safe and redundant. Snowball Edge also offers onboard encryption for security and AWS Greengrass integration, which could open doors for internet of things functionality. Greengrass supports local compute, messaging and data caching for connected devices. It also supports AWS Lambda functions, even when not connected to the internet.
Likewise, AWS Snowmobile, an Exabyte-scale data transfer service, carries the same advantages as Snowball Edge but at a much grander scale. Snowmobile can handle up to 100 petabytes, and it arrives in a tractor trailer.
Data transfer alternatives
The Snowball approach isn't the only way to transfer data to the public cloud. Another option is a hybrid cloud approach, in which raw data can be ingested, processed and summarized locally. IT teams can send smaller processed data sets to the cloud over networks for redundancy, sharing or even to take advantage of additional compute power in the public cloud.
There are other services to help businesses transfer data, ranging from laborious and manual to highly automated. Enterprises can use the Maestrano platform for a variety of data movement and integration tasks between cloud and non-cloud services, allowing data transfer instantaneously and automatically between systems. Alooma, a data pipeline company in Redwood City, Calif., also enables large wire movements -- as faster internet speeds cannot keep up with the increasing scale of data movement.
Even though Snowball and, in particular, Snowmobile allow vast movements of data, enterprises should start a large cloud move incrementally, and they should have a plan. And if a business needs to transform data on its way -- sometimes along the lines of traditional Extract, Transform and Load functions used for data warehouses -- it should look for a more suitable method, as that's a capability the Snowball still lacks.
Avoid these AWS migration mistakes
AWS' enterprise push continues with new tools
Snowball adds export functionality