AWS vs. Oracle cloud services: Compare costs and capabilities

The IaaS layer

AWS and Oracle clouds rely on compute, storage and networking resources to host users' virtual instances and scale workloads.

Core compute

Because AWS' and Oracle's computing architectures and chip types differ, we can't do a direct cost performance analysis. Both vendors have their own measures of vCPUs. AWS calibrates according to EC2 Compute Units, which relate to a standard measure of AWS machine performance. Oracle has standardized on Oracle Compute Units (OCPUs) for all its instance sizes.

AWS bundles its various VM services into different EC2 instance sizes. These sizes specify a certain number of vCPUs and amount of RAM. AWS adds storage to most instances with Amazon Elastic Block Store (EBS) or Amazon S3, which can store individual objects up to 5 TB in size. AWS users can aggregate an unlimited number of objects into an S3 bucket. Users can also configure some instance types to include local storage via low-cost hard drives or more high-performance solid-state drives (SSDs).

Oracle, on the other hand, bundles its VM services into instance sizes called VM shapes, which include a certain amount of OCPUs and memory. Oracle users usually purchase storage separately via a service called Block Volumes -- up to 1 petabyte. Users in need of high-performing local storage can also turn to DenseIO VMs, which work well for large databases. DenseIO variants can add 3.2 to 25.5 TB of local storage via attached SSDs that are connected via non-volatile memory express (NVMe) buses. Unlike AWS, Oracle does not offer a low-cost hard drive option.

A basic Oracle VM.Standard.E2.1 instance, with 1 OCPU, 8 GB of memory and 700 Mbps of networking -- at the time of publication -- costs $0.03 per hour. Oracle prices its instances in OCPU per hour, so an instance with 2 OCPUs, 16 GB of RAM and 1.4 Gbps of networking ends up being $0.06 per hour.

Both AWS and Oracle offer bare-metal instances that improve security and give users the ability to customize app infrastructure in the cloud. AWS calls its bare-metal offerings I3 instances. The I3 instance family starts with the i3.large with 2 vCPUs, 15.25 GB of memory and 475 GB of NVMe storage and costs $0.156 per hour. The largest one is currently the i3.metal, which includes 72 vCPUs, 512 GB of memory and 15.2 TB of NVMe storage and costs $4.992 per hour.

Oracle has several classes of bare-metal instances, starting with the BM.Standard.E2.64 with 64 OCPUs, 512 GB of RAM and 50 Gbps of aggregate networking throughput; it costs $0.03 per hour. Oracle has a DenseIO variant for big data workloads and large databases. The BM.DenseIO2.52 includes 52 OCPUs, 768 GB of memory and 51.2 TB of NVMe SSD storage and costs $0.1275 per hour.

Storage

Oracle's cloud storage pricing generally is lower than AWS', except for object storage services.

Oracle also offers users who need to transfer data out of the cloud a significant cost advantage over other cloud services. While data transfer into Amazon S3 or Oracle is free, it will cost you to transfer out. Data transfer out from AWS' S3 to the internet starts at $0.09 per GB after the first GB. Oracle outbound data transfers are free for the first 10 TB per month and then $0.0085 per GB.

Oracle's DenseIO instances have native NVMe storage and promise better performance for database applications. In contrast, AWS has a special class of EBS service for provisioned IOPS, which can be purchased as an add-on to EBS.

Networking

When it comes to the smaller machine sizes, AWS has the edge on raw networking capacity for cloud users. For example, its M5 instances start with 10 Gbps of network bandwidth and 3.5 Gbps of EBS bandwidth. Larger m5.large instances support up to 25 Gbps of bandwidth and 14 Gbps of EBS bandwidth.

At the low end, the comparable Oracle VM.Standard.E2.1 only comes with 700 Mbps network speed. However, networking capacity grows with the number of cores, up to an aggregate of 24.6 Gbps.

Feature set and security

Raw hardware power is far from the only consideration when enterprise IT teams compare AWS and Oracle cloud offerings. Public cloud services appeal to organizations on metrics such as availability and stability.

Availability

AWS was an early mover in the public cloud IaaS market. The reliability, availability and stability of its services reflect the maturity level of the AWS platform. AWS clearly leads in these categories, said Bill Saltys, senior vice president for alliances at Apps Associates, a cloud migration and managed services provider that works with both AWS and Oracle clouds.

By contrast, Oracle is relatively new to the public cloud market and, like all providers, must mature its offerings, delivery and support over time. The biggest challenge for all public cloud vendors is how fast they must go to not only keep pace with, but to gain any ground on, AWS, Saltys said.

One benefit for Oracle is that it offers end-to-end service-level agreements (SLAs) that cover performance, availability and manageability of services. Amazon provides SLAs as well, but these come with a long list of exclusions.

Depth of features

AWS outpaces all public cloud competitors when it comes to breadth of services and its feature set. The extent of AWS innovation and service release is tough to match. For example, the Los Angeles Clippers, a professional basketball team, invested in AWS technology, in part because AWS has augmented reality technology to transform the fan experience. These types of innovative features and capabilities are a core differentiator for AWS vs. Oracle and other cloud competitors.

Oracle has primarily focused on its own stack and database products, in terms of making them cloud-ready and available. This strategy makes sense and is something that Oracle must do in order to ensure a viable option and path forward for its customers and for the future of the company as a cloud computing provider, Saltys said.

For example, Oracle has delivered a set of robust SaaS cloud applications, which was its first as-a-service offering, followed by PaaS offerings in areas such as analytics and databases. However, when compared to AWS, Oracle and other providers do not cover the same landscape. Perhaps, there is no intention or reason to do so when it is more vital to deliver on cloud compute capabilities that align with the core mission of the company.

Governance

AWS infrastructure grew in response to startup success stories, such as Netflix and Airbnb, while Oracle has spent decades managing enterprise infrastructure. AWS' approach has made it easy for companies, especially startups, to quickly create new applications with its ecosystem of services. The downside is that this approach also makes it easier to leave glaring security holes, such as exposing sensitive data on unsecured S3 buckets. AWS has done a good job of implementing a robust identity and access management service.

In contrast, Oracle's approach to security arose from decades of building enterprise data management infrastructure, which had a strong governance and security need from the beginning. Oracle has created an architecture that makes it easy to separate application controls from data.

Oracle's approach considered the various ways that IT organizations leave security holes open, with more safeguards to prevent accidental data compromise. Development teams will have to decide for themselves whether these additional safeguards also slow the pace of app development lifecycles.

Hybrid and edge capabilities

In addition to public cloud services, AWS, Oracle and other major vendors are creating products to enable edge computing and hybrid cloud architectures.

Edge services

AWS builds most of its edge services to fit only its own ecosystem, while Oracle prefers to acquire smaller services that users can integrate into their development lifecycles.

Edge services refer to services that sit between the cloud and users, or even IoT products, like AWS IoT Greengrass. The former turn a collection of VMs into a robust web application framework. These include ancillary services like content distribution networks, firewalls, web proxies, monitoring and domain name system (DNS) services. AWS has branded various offerings, such as Amazon CloudFront for content distribution services, CloudWatch for monitoring and Route 53 for DNS services.

Most of these Amazon services were built to support the AWS ecosystem. This strategy could encourage developers to use the AWS tools and services they are familiar with when working on edge projects.

In contrast, to improve its core offerings, Oracle has been acquiring various companies, such as Dyn, a managed DNS service, and Zenedge, a hybrid security startup now called Oracle Dyn Web Application Security. Developers will have to evaluate how easy it is to work with these services.

Hybrid cloud

AWS has generally made it difficult for enterprises to create truly hybrid infrastructure. This scenario has led to the growth of several third-party hybrid cloud add-ons. Also, as noted above, Amazon S3 charges a substantial egress tax to transfer data out of the cloud. AWS has made some moves to simplify hybrid cloud provisioning, such as its partnership with VMware and launch of AWS Outposts that extends AWS infrastructure on premises.

As most of Oracle's existing business is on premises, the company pursues a strategy that enables users to migrate workloads into the cloud with security and governance baked in. As such, it offers better pricing to move data out of the cloud and has better integration options for enterprises' existing portfolio of Oracle applications.

In the AWS vs. Oracle cloud debate, a high-level evaluation of the specifications and costs of various infrastructure services suggests that Oracle has a lead in price-to-performance for many services. Oracle also aligns well with enterprise needs when it comes to security and hybrid applications. In the long run, the Oracle cloud will prove a viable option for enterprises with a large installed base of Oracle applications. Meanwhile, developers will have to decide for themselves the extent to which Oracle's development ecosystem represents a good alternative to AWS.