Oracle Exadata Database@AWS offers advantages, but trade-offs remain

Cloud giant allows customers to get more from Oracle hardware, but will try to gain more workloads on the side

DATABASES UPDATED As Oracle launches its Exadata service on Database@AWS — a year after it went live for lower-level databases — experts have warned there are trade-offs in getting value for money and cautioned over Oracle FUD around the cloud migration decision-making.

Earlier this month, Big Red announced the general availability of Oracle Exadata Database Service on Exascale Infrastructure on Oracle AI Database@AWS, which adds the beefed-up Oracle database to the service.

AWS was the last major cloud service to join the “database@” offer from Oracle, allowing the database and application biz to put its hardware inside the hyperscaler's datacenters.

Announced in 2024, Oracle Database@AWS can connect to applications running on Amazon Elastic Compute Cloud (Amazon EC2), AWS Analytics services, or AWS's advanced artificial intelligence (AI) and machine learning (ML) services, the vendor claims. The service became generally available in July last year.

Oracle now says it is bringing Exadata – the engineered hardware and software platform for high-performance databases — to AWS. The dominant cloud vendor has also signed a new long-term strategic collaboration agreement to help migrate customers.

Using Exadata allows customers to specify the compute and storage capacity they need, says Oracle, while spreading every database across pooled storage servers for high performance and availability, helping to avoid the burden of creating database and storage servers.

However, Oracle licensing and commercial experts have warned that there are still trade-offs when adopting the service and getting value for money.

While Database@AWS can be cheaper in terms of processors /CPU required than RDS, the compute and storage are more expensive and the service options are less granular, making it more difficult to right-size services, said Nick Walter, CTO and vice president of professional consulting services at House of Brick, an Oracle licensing advisory company.

“The Oracle Database@AWS offerings have been a constant topic of conversation with clients in the last year. It's a complicated topic because the cost structures, minimum configurations, and licensing metrics are so different from running Oracle databases in AWS native services such as EC2 or RDS,” he said.

For the database@AWS service, licensing metrics can be better than going to EC2, in that for users bringing their own licenses (BYOL), one processor license is required per 8 ECPU ratio, in contrast with the traditional one processor license per 2 vCPU ratio when running in RDS/EC2. Oracle's support rewards program can also offer good value.

However, compute and storage costs for Oracle Database@AWS offerings are “much higher than traditional RDS/EC2, and the capacity isn't as granular,” Walter said.

The Exadata offer helps with some of these challenges but does not eliminate them altogether.

“Exascale offerings may make the Oracle Database@AWS offering attractive to organizations with smaller fleets of Oracle databases as the provisioning is very granular,” he said.

While Exascale database storage costs per GB are competitive with the AWS Elastic Block Store GP3 options and offer higher I/O throughput, compute costs are higher than EC2/RDS, Walter said.

“Overall the Oracle Database@AWS offerings can be attractive to license-constrained organizations that need Oracle Database deployments in the cloud. Buyers should beware though, House of Brick has seen Oracle sales reps gleefully touting possible license savings and not mentioning the downsides,” he said.

They're all in this weird state of cooperation with Oracle, where now they're all authorized cloud providers. They're getting on stage together, and they can't say Oracle's crazy with their licensing

Craig Guarente, founder and CEO of Palisade Compliance, an Oracle licensing advisory company, said many organizations were still bogged down by legacy agreements for Exadata machines.

"Oracle has created some FUD, by saying they won't support customers on a particular cloud, [and] it's 'going to cost twice as many licenses if you just move to the generic cloud and not our services'. Oracle has created some artificial business blockers to get customers thinking about spending more money with them through these Exadata services,” he said.

Hyperscalers work with Oracle on these offers because they would love to get some of the business from Oracle on-prem workloads. Oracle can also create barriers to running their generic database in the cloud that might also get customers to consider OCI.

While promoting the database@AWS service, AWS is also telling customers to use its Bedrock AI to help move Oracle database applications to its PostgreSQL services.

“All of the hyperscale vendors — Google, Microsoft, AWS — are on one side, and Oracle on the other. They all want to generate more revenue; if they can take Oracle revenue and make it AWS revenue, I'm sure they would want to do that. However, they're all in this weird state of cooperation with Oracle, where now they're all authorized cloud providers. They're getting on stage together, and they can't say Oracle's crazy with their licensing. There's limited stuff that both parties can say, but I've definitely seen from the sales side — pick a vendor — that the knives are out,” Guarente said. ®

Updated to add on August 26:
An Oracle spokesperson has been in touch to say: “Oracle has partnered with AWS to offer Oracle Database in AWS in a variety of ways for well over a decade, letting our joint customers run Oracle Database alongside the applications they run in AWS. This gives AWS customers looking to run Oracle Database the added option to migrate to hardware that includes the integrated infrastructure for capabilities such as rapid failed-node detection and fast failover.

"Because Oracle controls the full stack in this configuration, from hardware to database software to cloud operations, the experience can be engineered, validated, and continuously improved end-to-end. This enables the high throughput utilization, ultra-fast response times, and extreme availability that RAC can deliver.

“It has been over a decade since Oracle and AWS began supporting Oracle Database together in AWS. Oracle and AWS recently announced that we signed a new Strategic Collaboration Agreement, and we are excited to continue to work with AWS to support our mutual customers as they continue to migrate their mission critical database workloads to the cloud.

“There are a wide range of service options for Oracle Database@AWS that include both dedicated and serverless infrastructure, giving customers the ability to choose shared storage and compute options at lower price points, or move to dedicated options that can scale to the largest workloads and maintain complete isolation,” they said.

In terms of the compute costs, the spokesperson said: “Oracle Autonomous AI Database Serverless offers zero infrastructure subscription costs, and Exascale Infrastructure starts at low CPU and storage amounts, and scales with customers’ needs. The offerings provide compelling TCO for customers of any size. The pricing for all of the Oracle Database@AWS products is also the same globally. While our joint offerings with AWS are already competitively priced in the lowest priced regions, they become even more so when compared to the higher prices many competing services present in some other regions.”