Cloud Economics
Oracle Cloud Infrastructure charges 50% less for compute, 70% less for block storage, 80% less for networking
Oracle Cloud Infrastructure (OCI) enables customers to save consistently on compute, storage, and networking compared with AWS, Azure, and Google Cloud. Even with complicated and overlapping discount programs, other clouds are still more expensive than regular OCI pricing.
Customer-friendly pricing
-
Lower-priced infrastructure
OCI infrastructure—compute, storage, and networking—delivers the same or better performance at a consistently lower price than other cloud providers. OCI offers a modern cloud experience unencumbered by the legacy design choices other cloud providers have made; we redesigned our hardware on top of bare metal servers and use off-box virtualization for the control plane. Choose to pay less with OCI.
-
A data-friendly network
Companies like yours send vast quantities of data across the globe to customers, partners, branch offices, recovery sites, and other cloud providers. Instead of holding your data hostage with exorbitant egress fees, OCI offers 10 TB of free data egress every month and up to 10X lower network charges than other providers. Move your data wherever you want.
-
Consistently low global pricing
OCI provides a consistent pricing experience in every region worldwide, including government regions. Where other providers have both higher and different prices in almost every region outside the US, OCI customers enjoy the same services, performance, and prices everywhere. Enable your global strategy while staying within budget.
-
Enterprise support included
OCI was purpose-built for mission-critical applications. The base fees for OCI services include enterprise-level support for those services. There is no extra charge for technical support for production workloads using OCI. This is in stark contrast with other cloud providers who can charge you 3% to 10% of the prior month’s (or year’s) bill, sometimes with a minimum fee, whether or not you ever contact support. Don’t pay extra for enterprise support.
-
Reduce your tech support bill
With Oracle Support Rewards, the more you use Oracle Cloud Infrastructure, the more you save. Customers can accrue US$0.25 to US$0.33 in rewards for every US$1 spent on OCI. Those rewards can be used to pay your on-premises tech software license support bill, even down to zero.1
-
Bring your own license
Keep your existing licenses for on-premises software at your negotiated rate when moving to supported OCI services or PaaS offerings. Use the Cost Estimator to model your savings.
-
Get help migrating
Oracle Cloud Lift Services provide guidance from cloud engineers on planning, architecting, prototyping, and managing cloud migrations. Clients can move critical workloads in weeks, or even days, instead of months by leveraging these included services for customer tenancies.2
-
Choose the compute you need
Unlike some other providers, OCI Compute allows you to scale the size of a virtual machine by a single CPU core and 1 GB/core on AMD, Arm-based, and Intel processors, so you avoid overpaying for unneeded capacity. Other providers can require you to double your compute size even when you need just a little more capacity.
Save significantly on foundational services
50% less
For compute
70% less
For block storage
80% less
For networking
Compute: compared to 2 OCPU (4 vCPU) AMD (E4) with 16 GB over the period of a month.
Block storage: compared to 1 TB, 25K IOPS, 240 MB/s throughput over the period of a month.
Networking egress over the public internet for 50 TB over the period of a month.
On-demand prices as of December 6, 2024 in the eastern US region.
OCI Cost Estimator
Estimate your monthly costs when using Oracle Cloud Infrastructure services.
Compute: The foundation of cloud services
Compute is at the heart of all cloud services and is the most well-known offering, whether virtual machines, containers, or serverless functions. OCI Compute is designed from the ground up to capitalize on bare metal performance and off-box virtualization, both of which provide predictable performance and avoid the noisy neighbor effect.
OCI Compute is one of the fastest and most cost-effective ways to migrate your existing applications to the cloud without re-architecting.
Compare compute costs across clouds
For a similar configuration on current hardware, OCI costs less.
57%
Less than AWS m6a.xlarge
57%
Less than Azure D4as v5
51%
Less than Google Cloud n2d-standard-4
Compared with the monthly cost of a two OCPU (four vCPU) AMD (E4) with 16 GB in US eastern regions. On-demand prices are as of December 6, 2024.
Monthly virtual machine cost
Virtual machines are a foundational service in the cloud, capable of running a variety of workloads. Virtual machines provide a high degree of control because you manage the operating system and any software you install.
The graph shows the costs of running a typically sized AMD-based virtual machine for an entire month. (Intel-based virtual machines show a similar trend.)
OCI charges the same in all regions, which is reflected in the graph. The other cloud providers charge a varying amount, depending on the region. The difference can be significant.
On-demand pricing—AMD, four vCPUs, 16 GB
Note: Google Cloud doesn’t offer government-only regions.
This bar chart shows a comparison of the cost (in US dollars) of a previous generation AMD-based compute virtual machine on OCI, AWS, Azure, and Google Cloud. A configuration of four vCPUs and 16 GB of memory is used for all comparisons.
Each bar on the chart represents the cost of running the virtual server for 730 hours, which is the length of an average month. Costs are compared for each of the following regions: eastern US, Brazil, UK, Germany, Singapore, Japan, and Australia. AWS doesn’t offer this generation of AMD compute in their government region, and Google doesn’t offer government regions.
The cost for OCI is about $54 in all regions, which is the lowest price offered by any of the hyperscalers in any region. AWS generally costs about $145, except for in Brazil, where the cost is about $200, and India, where the cost is about $81. Azure generally costs about $145, except for in Brazil, where the cost is about $200, and India, where the cost is about $81. The cost for Google Cloud generally ranges between $130 and $160, except for in India, where it’s about $65.
Monthly bare metal server cost
Bare metal servers allow you to leverage the full power of a server, including your own hypervisor, on OCI. Bare metal servers are ideal for high performance computing (HPC) workloads, where many servers operate as a clustered unit.
The graph shows the cost of running a bare metal server for an entire month. Similar to virtual machines, OCI charges the same for the same bare metal servers in all regions, which is reflected in the graph. The other cloud provider charges a varying amount, depending on the region.
On-demand pricing—AMD, 256 vCPUs, 2,048 GB
Note: Azure hasn’t publicly identified a replacement for their BareMetal series, and Google Cloud doesn’t publish pricing for their bare metal offering.
This bar chart shows a comparison of the cost (in US dollars) of a bare metal server on OCI and AWS.The bare metal server is based on an AMD processor. The server specifications for OCI are 128 OCPUs and 2,048 GB of memory. The server specifications for AWS are 192 vCPUs (which equate to 96 CPU cores) and 768 GB of memory.
Each bar on the chart represents the cost of running the bare metal server for 730 hours, which is the length of an average month. Costs are compared for each of the following regions: eastern US, Brazil, UK, Germany, Singapore, Japan, and Australia. Azure and Google don’t publish pricing for their bare metal offerings.
The cost for OCI is about $4,500 in all regions and is less than AWS in all regions. Generally, AWS costs about $7,000, except for in Brazil, where the price is almost $10,000.
Monthly GPU-enabled virtual machine cost
Graphical processing units (GPUs) enable you to perform many computationally demanding tasks, such as AI training and inferencing, data science activities, and computational fluid dynamics.
There are multiple GPUs available. This graph focuses on the NVIDIA A10 GPU to illustrate the clearest comparison between cloud providers. To provide the best performance, each cloud provider preconfigures the virtual machine to balance the CPU, memory, and GPU.
The graph shows the cost of running a preconfigured virtual machine with an NVIDIA A10 GPU for an entire month. OCI charges the same in all regions, whereas other cloud providers charge a varying amount depending on the region.
On-demand pricing—30 to 36 vCPUs, 128 to 440 GB
Note: Google Cloud doesn’t offer a virtual machine with the NVIDIA A10 GPU.
This bar chart shows a comparison of the cost (in US dollars) of a standard virtual machine with the NVIDIA A10 GPU on OCI, AWS, and Azure. Google Cloud doesn’t offer a virtual machine with the NVIDIA A10 GPU.
The virtual machine shape for OCI is VM.GPU.A10.1 and includes 15 OCPUs, 240 GB of memory, and one NVIDIA A10 GPU. The instance type for AWS is g5.8xlarge and includes four vCPUs (which equate to two CPU cores), 16 GB of memory, and one NVIDIA A10 GPU. The virtual machine type for Azure is NV36ads A10 v5 and includes 36 vCPUs (which equate to 18 CPU cores), 440 GB of memory, and one NVIDIA A10 GPU.
Each bar on the chart represents the cost of running the virtual machine for 730 hours, which is the length of an average month. The costs are compared for each of the following regions: eastern US, Brazil, UK, Germany, India, Singapore, Japan, Australia, and US Government. AWS and Azure don’t currently offer a virtual machine with an NVIDIA A10 GPU in their government region. AWS doesn’t currently offer an instance with an NVIDIA A10 GPU in Singapore.
The cost for OCI is about $1,500 in all regions, which is the lowest cost among all cloud providers. Generally, AWS costs about $2,300, except for in the eastern US region, where it’s about $1,800. Generally, Azure costs about $3,100, except for in the eastern US region, where it’s about $2,300, and in Brazil, where it’s about $4,600.
Monthly GPU-enabled bare metal instance cost
For AI and ML workloads that demand extreme performance, including clustering, OCI offers NVIDIA A100 80GB GPUs on bare metal instances. The comparison shows the closest comparison with other cloud providers, who offer only virtual machines for this configuration.
The graph shows the cost of running a preconfigured bare metal instance on OCI with eight NVIDIA A100 80GB GPUs for an entire month. OCI charges the same in all regions, whereas other cloud providers charge a varying amount depending on the region.
On-demand pricing—96 to 512 vCPUs, 1,152 to 2,048 GB
Note: The NVIDIA A100 80GB has limited availability.
This bar chart shows a comparison of the cost (in US dollars) of a standard bare metal or virtual machine with the NVIDIA A100 GPU with 80 GB of memory on OCI, AWS, and Azure. Only OCI offers the NVIDIA A100 GPU with 80 GB of memory on bare metal; all the other hyperscalers only offer it on a virtual machine.
The bare metal machine shape for OCI is BM.GPU.A100-v2.8. The virtual machine shape for AWS is p4de.24xlarge. The virtual machine shape for Azure is Standard_ND96amsr_A100_v4. The virtual machine shape for Google Cloud is a2-ultragpu-8g. The configurations vary from 96 to 512 vCPUs and from 1,152 GB to 2,048 GB of memory.
Each bar on the chart represents the cost of running the virtual machine for 730 hours, which is the length of an average month. The costs are compared for each of the following regions: eastern US, western US, Brazil, UK, Europe, India, Singapore, Japan, and Australia. AWS only offers their virtual machine in the US East and Singapore regions. Google Cloud only offers their virtual machine in the eastern US, western US, Europe, and Singapore regions.
The cost for OCI is about $23K in all regions, which is the lowest cost of all hyperscalers. AWS costs between about $30K and $37K. Azure generally costs about $30K, except for in Brazil, where the cost is about $48K. Google Cloud generally costs about $30K.
Flexible compute lets you save
OCI Compute allows you to match virtual machine resources to your workloads. You can scale by a single CPU core, which is equivalent to two vCPUs in other clouds, and you can scale memory in 1 GB increments per core. You can rightsize your performance—and cost—to meet your needs.
Other clouds provide compute in fixed sizes that typically double as they get bigger. Even if you need just a little bit more performance, you might have to double your compute size—and cost—to achieve it, which could result in significant overspending.
The graph shows the cost per hour of the latest AMD-based virtual machines at multiple sizes, from 2 to 96 vCPUs. All cloud providers have multiple sizes in the smaller range, but if you have a workload that needs 40 vCPUs, for example, you may have to purchase 48 vCPUs (AWS, Azure) or 60 vCPUs (Google Cloud) to achieve that performance. OCI lets you precisely select and scale your compute performance.
Per-hour virtual machine cost at multiple sizes
This line chart shows a comparison of the cost (in US dollars) of the latest AMD-based compute virtual machines on OCI, AWS, Azure, and Google Cloud. Customer workloads can vary in performance needs, which is why a range of vCPU sizes is shown, from 2 to 96 vCPUs.
OCI allows you to scale compute virtual machines in 2 vCPU increments, so the cost increases steadily as the size increases, resulting in a straight line on the chart.
AWS and Azure offer the following fixed sizes of virtual machines: 2, 4, 8, 16, 32, 48, and 64 vCPUs. The lines for AWS and Azure appear as a series of plateaus, as performance needs that fall between these fixed sizes require the next largest size. For example, a performance need for any size greater than 32 vCPUs but less than or equal to 48 vCPUs will require the same 48 vCPU size.
Google Cloud also offers fixed sizes for its latest AMD-based virtual machine. Its virtual machine sizes are 4, 8, 16, 30, 60, 90, and 180 vCPUs.
You can save with OCI
OCI offers globally consistent pricing, flexible consumption and deployment models, security services, and enterprise support included at no additional cost. Request a free cloud bill comparison today to see how much you could save by migrating some or all of your workloads to OCI.
Networking: The key to connecting cloud services
Networking enables cloud services to operate, communicate, and scale. OCI Networking allows you to create virtual data centers in the cloud that are securely isolated but fully connected to the entire range of OCI services.
Network design significantly impacts application performance. OCI created a nonblocking network that provides performance guarantees.
Compare data egress costs across clouds
For 50 TB of data egress, OCI costs less.
92%
Less than AWS
90%
Less than Azure standard tier
90%
Less than Google Cloud
Compared with the monthly cost of outbound data leaving US regions. Prices are as of December 6, 2024.
Monthly data egress fees
Public internet
A per-byte fee is charged for data leaving a region—a data egress cost. Because the networking infrastructure is owned by different companies, the pricing varies across the globe for all cloud providers.
However, the other cloud providers can use data egress fees to trap your data by making it too expensive to move data outside of one of their regions.
OCI is different. OCI includes 10 TB of monthly data egress for free, which is significantly more than the other cloud providers. After the free 10 TB, OCI charges significantly less. This graph shows how your costs can escalate as you move more data from different cloud providers’ global regions to a US destination.
This bar chart shows a comparison of the cost (in US dollars) of data movement from a cloud provider’s region to an external destination over the internet. Where there are multiple tiers of public internet access, the standard or least expensive tier has been selected.
Each bar on the chart represents the cost of moving a certain amount of data within a single month. The chart shows how the cost of moving 50 TB, 100 TB, and 500 TB of data in Europe, South America, and Japan differs based on the cloud provider.
OCI is significantly less expensive than the other cloud providers in all regions.
The cost of moving 50 TB from Europe is about $340 with OCI, $4,300 with AWS, $3,400 with Azure, and $3,400 with Google Cloud. The other cloud providers are at least 10X more expensive than OCI.
The cost of moving 50 TB from South America is $1,000 with OCI, $7,000 with AWS, $4,600 with Azure, and $4,600 with Google Cloud. The other cloud providers are about 5X more expensive than OCI.
The cost of moving 50 TB from Japan is $1,000 with OCI, $4,700 with AWS, $4,100 with Azure, and $4,100 with Google Cloud. The other cloud providers are at least 4X more expensive than OCI.
The cost of moving 100 TB from Europe is about $800 with OCI, $7,800 with AWS, $6,400 with Azure, and $6,700 with Google Cloud. The other cloud providers are at least 8X more expensive than OCI.
The cost of moving 100 TB from South America is $2,200 with OCI, $13,000 with AWS, $8,600 with Azure, and $8,800 with Google Cloud. The other cloud providers are about 4X more expensive than OCI.
The cost of moving 100 TB from Japan is $2,200 with OCI, $9,000 with AWS, $7,600 with Azure, and $7,900 with Google Cloud. The other cloud providers are at least 3X more expensive than OCI.
The cost of moving 500 TB from Europe is about $4,200 with OCI, $29,000 with AWS, $23,000 with Azure, and $26,000 with Google Cloud. The other cloud providers are at least 6X more expensive than OCI.
The cost of moving 500 TB from South America is $12,000 with OCI, $60,000 with AWS, $39,000 with Azure, and $41,000 with Google Cloud. The other cloud providers are at least 3X more expensive than OCI.
The cost of moving 500 TB from Japan is $12,000 with OCI, $43,000 with AWS, $32,000 with Azure, and $36,000 with Google Cloud. The other cloud providers are at least 3X more expensive than OCI.
Private line, 10 Gb/sec
If you need more control or consistent connectivity, you can use a private line service. A private (or dedicated) line can have two pricing components: a “port” charge and a per-byte charge.
OCI is again different because there is no per-byte charge. All the other cloud providers have both a port and per-byte charge.
For a dedicated, 10 Gb/sec connect, the pricing difference is significant. The OCI FastConnect service is just US$931 per month, regardless of how much data is transferred. The other cloud providers start charging significantly more when the data volume increases.
The pricing doesn’t include the third-party cost of the private line itself, which varies globally.
Note: Azure costs are based on the best metered or unlimited prices.
This bar chart shows a comparison of the cost (in US dollars) of data movement from a cloud provider’s region to an external destination over a 10 Gb/sec dedicated connection.
Each bar on the chart represents the cost of moving a certain amount of data within a single month. The chart shows the differences in cost across cloud providers for moving 30% of the available bandwidth, which is 986 TB; 50% of the available bandwidth, which is 1,643 TB; and 75% of the available bandwidth, which is 2,464 TB. The source regions are Europe, South America, and Japan. The destination region for all of them is the US.
OCI is dramatically less expensive than the other cloud providers in all regions because it only charges a per-hour port fee for a dedicated connection. OCI doesn’t charge for bandwidth consumption.
For all scenarios, OCI charges about $900.
The cost of moving 986 TB from Europe is $29,000 with AWS, $28,000 with Azure, and $21,000 with Google Cloud. The other cloud providers are at least 22X more expensive than OCI.
The cost of moving 986 TB from South America is $149,000 with AWS, $82,000 with Azure, and $110,000 with Google Cloud. The other cloud providers are at least 90X more expensive than OCI.
The cost of moving 986 TB from Japan is $90,000 with AWS, $53,000 with Azure, and $43,000 with Google Cloud. The other cloud providers are at least 46X more expensive than OCI.
The cost of moving 1,643 TB from Europe is $48,000 with AWS, $44,000 with Azure, and $35,000 with Google Cloud. The other cloud providers are at least 37X more expensive than OCI.
The cost of moving 1,643 TB from South America is $248,000 with AWS, $82,000 with Azure, and $182,000 with Google Cloud. The other cloud providers are at least 88X more expensive than OCI.
The cost of moving 1,643 TB from Japan is $149,000 with AWS, $82,000 with Azure, and $71,000 with Google Cloud. The other cloud providers are at least 76X more expensive than OCI.
The cost of moving 2,464 TB from Europe is $71,000 with AWS, $51,300 with Azure, and $51,000 with Google Cloud. The other cloud providers are at least 55X more expensive than OCI.
The cost of moving 2,464 TB from South America is $371,000 with AWS, $82,000 with Azure, and $272,000 with Google Cloud. The other cloud providers are at least 88X more expensive than OCI.
The cost of moving 2,464 TB from Japan is $223,000 with AWS, $82,000 with Azure, and $105,000 with Google Cloud. The other cloud providers are at least 88X more expensive than OCI.
Speedy storage: High I/O rate and dynamic performance
Some workloads require demanding I/O: high operations per second and high throughput bandwidth. OCI offers flexible, NVMe-based block storage volumes with the ability to select your desired performance—and only OCI provides a performance SLA for block storage.
You can change the performance of OCI Block Volumes dynamically without detaching or recreating them. You can also enable automatic tuning so the performance of the volumes adjusts between limits you set—all at competitively low prices.
Compare block storage costs across clouds
For a similar configuration on current hardware, OCI costs less.
82%
Less than AWS
gp3, 100 GB, 6K IOPS
82%
Less than Azure
Premium SSD v2, 100 GB, 6K IOPS
75%
Less than Google Cloud
Hyperdisk Balanced, 100 GB, 6K IOPS
Compared with OCI block storage, 100 GB, balanced performance, 6K IOPS. Prices are as of December 6, 2024.
Monthly block storage cost
On-demand pricing for balanced performance, from 6K to 25K IOPS
Block storage is the performant storage you attach to virtual machines or bare metal servers. In addition to size, you can select performance characteristics, such as the throughput (measured in MB/sec) and I/O operations (measured in input/output operations per second (IOPS)).
For moderate workloads, a balanced performance is sufficient, which means lower performance characteristics.
The graph shows the cost of block storage that provides 6K IOPS. If the cloud provider offered multiple options, including the need to purchase “performance credits,” the cheapest option was selected. Additional options, such as replication, were declined. For AWS, the choice was gp3. For Azure, it was Premium SSD v2. For Google Cloud, it was either Hyperdisk Balanced or Extreme Persistent Disk.
OCI provides the needed performance at a significantly lower cost for the selected sizes.
This bar chart shows a comparison of the cost (in US dollars) of block storage that can be attached to a virtual machine. The comparison was performed using costs for the eastern US region, where the pricing is lowest.
For the comparison, the block storage need only provide balanced performance for typical workloads, which results in throughput between 6K to 25K input/output operations per second. The sizes of the block storage are 500 GB, 1,000 GB, and 10,000 GB. In all cases, the block storage was configured to provide at least the same throughput as OCI Block Storage, if it didn’t automatically match the throughput at the desired size.
For OCI, OCI Block Storage is used. For AWS, gp3 EBS is used. For Azure, Premium SSDv2 is used. For Google Cloud, the cheaper of Hyperdisk Balanced or Extreme Persistent Disk is used.
In all scenarios, the other cloud providers are significantly more expensive than OCI.
For 500 GB in block storage, OCI costs $21, AWS costs $155, Azure costs $160, and Google Cloud costs $154. The other cloud providers are at least 7X more expensive than OCI.
For 1,000 GB in block storage, OCI costs $43, AWS costs $195, Azure costs $200, and Google Cloud costs $187. The other cloud providers are at least 4X more expensive than OCI.
For 10,000 GB in block storage, OCI costs $430, AWS costs $920, Azure costs $930, and Google Cloud costs $910. The other cloud providers are at least 2X more expensive than OCI.
On-demand pricing for high performance, from 8K to 300K IOPS
Block storage needs to provide significantly more performance for high performance workloads, such as a database or HPC. With large workloads, the performance demands can escalate significantly as the size increases.
The graph shows the cost of block storage that provides increasing throughput, from 38K IOPS up to a blistering 300K IOPS, as the storage size increases.
Again, the cheapest option was selected for each cloud provider and any additional options were declined. For AWS, the choice was io2, as gp3 tops out at just 16K IOPS. For Azure, it was Premium SSD v2, except for the most demanding use case, where Ultra Disk was chosen. For Google Cloud, it was Hyperdisk Balanced, except for the most demanding use case, where Hyperdisk Extreme was chosen.
OCI again provides the needed performance at a significantly lower cost for the selected sizes.
This bar chart shows a comparison of the cost of block storage that can be attached to a virtual machine. The comparison was performed using costs for the eastern US region, where the pricing is lowest.
For the comparison, the block storage must provide high performance for demanding workloads, such as a database or cluster software. This results in throughput starting at about 40K input/output operations per second for a 500 GB volume, going up to 75K for a 1,000 GB volume, and topping out at 375K for a 5,000 GB volume.
For OCI, OCI Block Storage is used. For AWS, io2 EBS is used to achieve the higher IOPS. For Azure, the cheaper of Premium SSDv2 or Ultra Disk is used. For Google Cloud, the cheaper of Hyperdisk Balanced or Hyperdisk Extreme is used.
In all scenarios, the other cloud providers are dramatically more expensive than OCI.
For 500 GB in block storage at 38K IOPS, OCI costs $30, AWS costs $2,400, Azure costs $220, and Google Cloud costs $270. The other cloud providers are at least 8X more expensive than OCI.
For 1,000 GB in block storage at 75K IOPS, OCI costs $60, AWS costs $4,000, Azure costs $460, and Google Cloud costs $540. The other cloud providers are at least 8X more expensive than OCI.
For 10,000 GB in block storage at 375K IOPS, OCI costs $300, AWS costs $10,000, Azure costs $19,000, and Google Cloud costs $12,000. The other cloud providers are at least 35X more expensive than OCI.
Containers and serverless: Enabling cloud native development
Cloud native architectures and modern technologies (including microservices, containers, Kubernetes, and serverless) are transforming the way we design, develop, and ship applications. OCI provides rapid time to value when building modern apps.
OCI Kubernetes Engine (OKE) enables you to simplify the operation of enterprise-grade Kubernetes as a managed service. Serverless Kubernetes allows you to scale your containerized application without manually administrating the clusters.
Compare serverless Kubernetes costs across clouds
OKE provides more value for less money for serverless operations.
68%
Less than Amazon EKS with AWS Fargate
68%
Less than Azure AKS with virtual nodes
70%
Less than Google Cloud GKE Autopilot
Compared with OKE virtual nodes, the monthly cost of a 20-pod cluster on serverless Kubernetes with 16 vCPUs and 64 GiB of memory per pod. Prices are as of December 13, 2024.
Monthly pricing for serverless Kubernetes on x86
Kubernetes provides sophisticated resource management, including the ability to specify CPUs and memory limits for containers. Kubernetes can help you simplify operations with automation, save time on infrastructure management, increase resource utilization and efficiency, and improve agility, flexibility, uptime, and resilience.
The graph shows the monthly cost of serverless Kubernetes on x86 with Linux using a 20-pod cluster with 16 vCPUs and 64 GiB of memory per pod.
This bar chart shows a comparison of the cost (in US dollars) of a serverless Kubernetes cluster on OCI Kubernetes Engine (OKE) virtual nodes, Amazon Elastic Kubernetes Service (EKS) with AWS Fargate, Azure Kubernetes Engine (AKS) virtual nodes, and Google Cloud Kubernetes Engine (GKE) Autopilot.
The cluster consists of 20 pods, each with 16 vCPUs and 64 GiB of memory per pod, using x86 architecture with Linux. Each bar on the chart represents the cost of running the cluster for 744 hours, which is the length, in hours, of 31 days.
OCI prices are consistent regardless of the region, but the other options vary quite widely in price depending on the region. For the comparison, we’ve used the lowest published regional rates for AWS, Azure, and Google to show them at their best.
Prices reflect each service’s recommended options for production workloads, such as Standard tier for AKS, Standard edition for GKE, and three virtual nodes for OKE. For simplicity, we’ve also assumed the use of general purpose processors with the included ephemeral storage allocation. GKE offers a free tier providing $74.40 in monthly credits, which we have applied.
We have attempted to use the closest comparisons in terms of the Kubernetes version and processor configurations, such as manufacturer, type, generation, speed, memory, bandwidth, and so on.
Common cloud industry practice is to define compute instances based on the number of virtual CPUs they include. Each vCPU provides the capacity for one thread of runs. A vCPU doesn’t provide a whole physical compute core but is instead part of a core. In contrast, Oracle’s x86 compute shapes use OCPUs, which equate to physical CPU cores, each of which provides for two threads. So two x86 vCPUs on AWS, Azure, or Google Cloud are comparable to one OCPU on OCI.
The cost of OKE virtual nodes is about $4,500 in all regions, which is far less than the cost of the other three options. The cost of both EKS with Fargate and AKS virtual nodes is about $14,000, and the cost of GKE Autopilot is more than $15,000.
Monthly pricing for serverless Kubernetes on Arm
Arm processors are significantly more power-efficient than x86, often making them the preferred choice for cost-effective containerized applications orchestrated by Kubernetes. Both Arm and x86 are available with OKE, allowing you to choose the one that best suits your specific needs.
The graph shows the monthly cost of serverless Kubernetes on Arm with Linux using a 20-pod cluster with 16 vCPUs and 64 GiB of memory per pod.
Note: Azure AKS virtual nodes don’t offer support for Arm shapes.
This bar chart shows a comparison of the cost (in US dollars) of a serverless Kubernetes cluster on OCI Kubernetes Engine (OKE) virtual nodes, Amazon Elastic Kubernetes Service (EKS) with AWS Fargate, Azure Kubernetes Engine (AKS) virtual nodes, and Google Cloud Kubernetes Engine (GKE) Autopilot.
The cluster consists of 20 pods, each with 16 vCPUs and 64 GiB of memory per pod, using Arm architecture with Linux. Each bar on the chart represents the cost of running the cluster for 744 hours, which is the length, in hours, of 31 days. Note that Azure AKS virtual nodes don’t offer support for Arm shapes, hence there is no bar in the chart for AKS virtual nodes.
OCI prices are consistent regardless of the region, but the other options vary quite widely in price depending on the region. We’ve used the lowest published regional rates for AWS and Google to show them at their best.
Prices reflect each service’s recommended options for production workloads, such as Standard edition for GKE and three virtual nodes for OKE. For simplicity, we’ve also assumed the use of general purpose processors with the included ephemeral storage allocation. GKE offers a free tier providing $74.40 in monthly credits, which we have applied.
We have attempted to use the closest comparisons in terms of the Kubernetes version and processor configurations, such as manufacturer, type, generation, speed, memory, bandwidth, and so on. Arm vCPUs and OCPUs are comparable across all providers: One Arm vCPU equals one Arm OCPU.
The cost of OKE virtual nodes is about $4,000 in all regions, which is far less than the cost of the other two options. The cost of EKS with Fargate is more than $11,000, and the cost of GKE Autopilot is more than $12,000.
Distributed cloud: Put the cloud where you need it
Oracle Cloud Infrastructure’s distributed cloud offers customers the flexibility to deploy their workloads wherever they like—even across multiple clouds—with the benefits of cloud innovation and greater control over data residency, locality, and authority. Customers use OCI’s distributed cloud to satisfy their business, regulatory, and performance requirements, which are often not met by other public cloud providers.
OCI pricing is consistent across all regions
Other cloud providers charge higher prices in different regions for the same services.
AWS
59% higher
in the AWS São Paulo region than in the US East region for the same AWS m6a.xlarge instance type
Azure
60% higher
in the Azure São Paulo region than in the East US region for the same Azure D4as v5 virtual machine type
Google Cloud
41% higher
in the Google Cloud São Paulo region than in us-east1 for the same Google Cloud n2d-standard-4 VM
On-demand prices are as of December 6, 2024.
Percentage price increase over a US region for the same VM type
OCI is unique among the cloud providers in that it provides the same services at the same prices in all regions. This provides customers with the confidence to add or move workloads as needed without encountering unexpected expenses.
The other cloud providers can charge significantly more for the same service. The graph shows how the price of the same virtual machine can vary dramatically from the price in the provider’s US eastern regions. For example, both AWS and Azure charge almost 60% more for the same virtual machine in the Brazil region.
This bar chart shows a comparison of the cost (in US dollars) of a standard virtual machine in the eastern US region and non-US regions, including Brazil, the UK, Germany, Singapore, Japan, and Australia.
Typically, any service offered by the other cloud providers is cheapest in the US regions. Other cloud providers can charge significantly more for the same service in non-US regions.
OCI charges the same rate for each service in all regions, which makes for consistent pricing and budgeting.
For the same size of virtual machine, AWS and Azure charge 60% more in Brazil than in the eastern US region. Google Cloud charges 40% more.
In the UK, Germany, Singapore, Japan, and Australia, AWS and Azure charge about 20% more than in the eastern US region. With the exception of Australia, Google Cloud is about 15% more expensive in these regions than in the eastern US region. In Australia, Google Cloud is about 20% more expensive.
Government cloud: The same price as commercial
OCI Government Cloud—for IaaS and PaaS—is priced the same as our commercial public cloud and meets U.S. Defense Department (Impact Levels 2 and 4) and FedRAMP High authorization standards. Oracle also operates a sovereign, dual-region service in the UK for UK government and defense customers.
Commercial customers who need to meet government accreditation requirements should consider using OCI Government Cloud, which can enable them to achieve their compliance goals.
| Service | OCI SKU | OCI hourly cost | AWS SKU | AWS hourly cost | Azure SKU | Azure hourly cost |
|---|---|---|---|---|---|---|
| Virtual machine, 4 vCPUs, 16 GB | VM.Standard.E4.Flex | $0.07 | D4as v5 | $0.22 | ||
| Storage optimized virtual machine, 16 vCPUs, 120 GB | VM.DenseIO.E4.Flex | $0.85 | i3.4xlarge | $1.50 | L16s v3 | $1.65 |
| Bare metal 128 vCPUs, 1024 GB | BM.Standard3.64 | $4.10 | m6i.metal | $7.74 | ||
| Managed Kubernetes, 192 vCPU, 1,536 GB | VM.Standard.E4.Flex | $4.80 | r5a.16xlarge | $13.16 | E64as v5 | $13.16 |
| Block storage, 1 TB, 1 volume, 15K IOPS, 125 MB/sec | Block storage (balanced) | $0.06 | EBS gp3 | $0.28 | Premium SSD v2, LRS | $0.29 |
| Data egress, 50 TB | First 10,000 GB free | $0.47 | First 100 GB free | $5.88 | First 100 GB free | $4.65 |
| Private line, 10 Gb/sec, 326 TB (10% of bandwidth) | FastConnect circuit fee | $1.28 | Direct Connect | $11.25 | Express Route | $17.07 |
All prices assume 730 hours per month and constant usage throughout the month. Prices rounded to the nearest cent.
Try Oracle Cloud today.
Discounting made simple
-
OCI
Oracle offers Universal Credits, which can be used for IaaS and PaaS services across all regions. You aren’t restricted to a particular compute type or service, and you don’t need to specify your allocations in advance.
If your needs change, if your workloads shift, or if new services are offered, you can switch and still use your Universal Credits for a discount.
You can commit to one or three years; increased commitment levels result in higher discounts. If you use up your Universal Credits before the end of the term, you still enjoy the discounted service rates for the rest of your term.
OCI also offers reserved capacity, provisioned concurrency, preemptible instances, and more. And you can bring your Oracle licenses from on-premises (PDF).
-
AWS
AWS offers a commitment-based discount, known as an Enterprise Discount Program (EDP). AWS also offers reserved instances, an EC2 discount savings plan, a compute discount savings plan, a SageMaker discount savings plan, and spot instances.
Not all discount savings plans are compatible with an EDP, which means you have to estimate your usage over the next year or three years if you want to take advantage of them.
What if your needs change? If you commit to a savings plan, you’re locked in. Some discount programs are tied to a specific region, requiring you to guess which region will be best. If a new region opens up, your discount program might not be applicable there.
If you’re bringing your Oracle Database licenses, you get twice as many CPUs on OCI as you do on AWS (PDF).
-
Azure
Azure offers a commitment-based discount, known as an Enterprise Agreement (EA). Azure also offers reserved instances, a savings plan for compute, Azure Hybrid Benefit, and spot instances.
The savings plan for compute isn’t listed as compatible with an EA, which means you have to estimate your usage over the next year or three years if you want to take advantage of it.
What if your needs change? If you commit to the savings plan, you’re locked in. Or, if you choose reserved virtual machines, you’re tied to a specific region, requiring you to guess which region will be best. If a new region opens up, your reserved virtual machine won’t be transferable.
If you’re bringing your Oracle Database licenses, you get twice as many CPUs on OCI as you do on Azure (PDF).
-
Google Cloud
Formally, Google Cloud doesn’t advertise an enterprise discount that applies to all services. Google does state in their financial filings that they do offer discounts to customers, but they don’t disclose if these discounts are part of a program or custom discounts offered per customer.
Google Cloud also offers an automatic discount (sustained use discount) and two forms of a commitment-based discount (committed use discounts). These discounts can’t be combined with each other and may or may not be able to be combined with an enterprise agreement. You have to estimate your usage over the next year or three years if you want to take advantage of committed use discounts.
What if your needs change? If you commit to the committed use discount, you’re locked in. For some services, the committed use discount is tied to a specific region, requiring you to guess which region will be best. If a new region opens up, your discount might not be applicable there.
1 Paying down the technology support bill is subject to regional tax regulations. Taxes can’t be paid with Support Rewards. Only the pretax invoice amount can be paid with Support Rewards.
2 Cloud Lift Services are available in many regions. Contact your account team for details and availability.
Unless otherwise stated, all prices were current as of December 6, 2024.