chris - Fotolia

Amazon Aurora expands reach, but maybe not far enough

Support for PostgreSQL in Amazon Aurora helps AWS advance in its fight with Oracle over a larger piece of the database market. But it does not offer total feature parity.

Amazon has taken another big step in its quest to lure companies to swap out Oracle databases for AWS, but it can't...

remove every roadblock that might prevent IT departments that look to switch.

In recent weeks, AWS fulfilled a pledge from a year ago when it added PostgreSQL support in Amazon Aurora. Many cost-cutting corporations view PostgreSQL, also known as Postgres, as the best open source alternative to Oracle, and its inclusion in Aurora means AWS now can handle most, but not all, of these workloads.

Amazon Aurora is a commercial version of an internal Amazon database built on MySQL, another open source database. AWS executives have said it's the fastest growing service in the company's history. Amazon said this new PostgreSQL flavor of Aurora runs three times faster than standard PostgreSQL, and industry observers said the added support from AWS makes it suitable for a wide range of uses.

Open source PostgreSQL covers about 80% of what is found in Oracle's relational database, and Amazon Aurora addresses about 80% of features PostgreSQL doesn't address in terms of management, monitoring and high availability, said Jim Mlodgenski, CTO of OpenSCG. The AWS consulting partner based in East Brunswick, N.J., has done Oracle-to-PostgreSQL conversions for 15 years.

If I'm looking at new applications or starting a new company then I would certainly consider these databases … but not if I had a huge legacy footprint.
Tony Baeranalyst, Ovum

"It's kind of a game-changer when it comes to Postgres to push it that last mile to major enterprises [that] need that high availability and monitoring that open source Postgres is still working on," he said.

Most enterprises use cloud databases, such as Aurora, for new application development, though Amazon hopes corporations will port over existing workloads in time. AWS has several features, such as AWS Database Migration Service and AWS Schema Conversion Tool, which help transition from legacy systems.

While those features can help customers get up and running on AWS, Aurora lacks a sliver of high-end functionality that's just enough to dissuade heavy Oracle users, said Kevin Epstein, CTO at Onica, an AWS consulting partner in Santa Monica, Calif.

"You may be lucky and have not embraced some features and functionality that Oracle offers, but the reality is most people bought those [services] because of the secret sauce it offers," Epstein said.

Tony BaerTony Baer

There are also practical reasons not to swap Oracle for Amazon Aurora, regardless of how well the service runs with PostgreSQL. Because the release is so new, it is really only suitable for a pilot or stand-alone project, said Tony Baer, an analyst with Ovum, based in London.

"Postgres is designed to be an enterprise, robust database, but it's not going to match [Oracle] PL/SQL word for word on the structure. So, even if it were at the 3.0 release right now, I would say don't unplug your database and shift overnight," Baer said.

Legacy systems with hundreds of thousands of lines of code written directly into the Oracle database probably aren't worth the cost or effort to move over, industry observers said. Amazon Aurora also lacks some of the enterprise-grade attributes that can be found in Oracle, which include levels of governance, security, backup and recovery.

"The number of features that Oracle built over the decades around some of these big compression pieces is not there; some data warehouse-type features are still not there," Mlodgenski said. "But those come up so rarely -- not company by company, but application by application."

Most enterprises Mlodgenski deals with won't completely abandon Oracle, but they've started to transition some of the more obvious applications to at least curb licensing costs. OpenSCG helps customers migrate apps to private data centers and into AWS or Microsoft Azure. In the past year, the share of cloud migrations jumped from about a third to three-quarters of their total jobs. Part of that is the maturity of PostgreSQL in Amazon Relational Database Service. But another reason is that customers are more comfortable with cloud computing, Mlodgenski said.

Cloud databases have the edge for new development

Most AWS database services involve building something new in the cloud instead of migrating workloads from private data centers -- though the latter is starting to kick into gear, said Merv Adrian, an analyst with Gartner. Cloud-based databases can be more appealing than traditional counterparts for newer mobile and digital applications with global presences thanks to capabilities such as inexpensive capacity and automatic scalability.

And while they may not be as good as transactional databases for general ledgers or accounts payable, that's not where growth lies in the database market.

"If I'm looking at new applications or starting a new company, then I would certainly consider these databases," Baer said. "But not if I had a huge legacy footprint; not if I was a Global 2,000 corporation trying to move over transactional systems."

AWS' database revenues grew by 108% in 2016 to $1.7 billion, but that's dwarfed by Oracle which brought in $13.9 billion in that same time frame, according to Gartner estimates. Oracle won't cede the top spot in the market any time soon as annuities on existing workloads will churn revenue for years to come, Adrian said.

But Oracle does face a real threat for next-generation applications better suited for the cloud and the developers that want to work there. Amazon is increasingly the alternative for a newer generation of programmers, particularly those who work within lines of business such as marketing where there might be less red tape needed to make a purchase, Adrian said.

Of course, IT shops can run Oracle on AWS, which offloads some maintenance duties, but that rarely translates to the cost savings that attracted many companies to the cloud in the first place. In fact, it may be more expensive after Oracle changed its licensing policies this year, which essentially double the fees for running on AWS.

And Oracle isn't AWS' only competition. Microsoft is the No. 2 database retailer at $7 billion in revenue in 2016, according to Gartner, and also is the No. 2 public cloud provider. And Microsoft has just begun to directly compete with Oracle with support for Linux in SQL Server. Also this year, Google Cloud Spanner and Azure Cosmos DB added new wrinkles to the concept of a globally distributed database, though AWS will likely respond soon with improved distribution for Amazon Aurora across regions, Adrian said.

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

Next Steps

AWS targets data lake migrations

Amazon Relational Database Service offers diverse database options

Azure jumps into cloud database fray

Dig Deeper on AWS database management