Kit Wai Chan - Fotolia

AWS database migration push continues with data lake conversions

New AWS database migration capabilities simplify data lake transfers to Redshift from Oracle and Teradata, but it still doesn't solve the biggest obstacles to move databases to the cloud.

AWS continues to find ways to make it easier to move data off Oracle and onto its cloud -- while heating up competition between the two companies.

AWS database migration capabilities have been updated via its Schema Conversion Tool (SCT) to make it easier to export data lakes out of Oracle and Teradata and move them into Amazon Redshift. This comes on the heels of Oracle's decision last month to hike licensing fees for new customers using its databases on AWS or Microsoft Azure.

AWS SCT analyzes a company's data warehouse, automates the schema conversion between the environments and extracts the data, regardless of volume. Customers can use Amazon Simple Storage Service or Amazon Import/Export Snowball to export the data to AWS.

Boingo Wireless Inc., a Los Angeles-based mobile internet access provider that uses Oracle and AWS, migrated its BI databases from Oracle to Redshift last year. The physical AWS database migration didn't take long because of the size of the workload and because of Redshift's compression capabilities, but there was considerable effort to select a third-party tool for the schema conversion, said Kishore Raja, vice president of engineering and strategic programs at Boingo.

"We didn't have dozens of terabytes of data to move, in which case it would have been really awesome to [have this new] automation," he said. "Still, if we had this tool back in 2014 to 2015 we definitely would have explored it rather than third-party tools or other tools we were taking a lot of time on."

Most data warehousing applications integrate with business intelligence tools that support AWS, so pointing to the new platform should go relatively smoothly if their application is supported by Redshift and Oracle, said Adam Ronthal, research director at Gartner.

"Data warehousing is a good place to start," he said. "It's a lot less complicated than transactional or operational databases."

Schema and data conversions are the easiest part of AWS database migration, Ronthal said. What's difficult -- and what AWS still hasn't solved in a meaningful way -- is application portability, especially for custom code that must be rewritten.

"If I've got a thousand lines of SQL written in an Oracle back end I need to convert to run on MySQL or PostgreSQL, there's a lot of work involved and it's not push-of-a-button and it won't go as easy as [AWS] might want to make it look," Ronthal said.

The other option is for customers to run Oracle directly on AWS instances or through Amazon Relational Database Service, but a recent change by Oracle that effectively doubled processor licensing requirements on AWS could make it cost prohibitive.

Databases central to cloud competition

Databases -- and the inherent workload gravity that comes with them -- have become a huge part of the scrum between cloud vendors, as analytics-heavy startups continue to scale and churn through data and as enterprises contemplate the shift from on premises.

Amazon is determined to expand its database presence and increase the means to migrate databases to its platform. This new SCT capability joins a list of similar capabilities under the AWS Database Migration Service launched in late 2015. According to Amazon, more than 14,000 databases have been migrated to AWS since the service became available.

"In and of itself [this new capability] is incremental," said Tony Baer, principal analyst at Ovum based in London. "It's part of a steady strategy on AWS' part to lower the obstacles to making the idea of migrating legacy databases Aurora more thinkable."

Aurora is AWS' fastest growing service. The fully managed product is MySQL-like, and AWS late last year added support for PostgreSQL, which is the open-source option that most closely resembles Oracle.

Much of the public cloud adoption has so far focused on test and development or to shift from Capex to Opex, but the coming trend will be for managed services that remove management of patches, updates and configuration, Baer said. That keeps the focus on the database management, but it also opens the doors to all the services that hook into Aurora.

"You start using other services [that link to Aurora] like Kinesis or Lambda and you're using services that are uniquely Amazon, so you're basically buying a new vendor platform," Baer said.

Trevor Jones is a news writer with SearchCloudComputing and SearchAWS. Contact him at [email protected].

Next Steps

Amazon Redshift has its strengths and quirks

Here's how to approach a migration from Oracle to AWS

What to know about moving proprietary databases to AWS

Dig Deeper on AWS database management