The latest cloud infrastructure announcements, technical solutions, and enterprise cloud insights.

What Everyone Should Know About Cloud Consumption

Ed Beauvais
Director, Product Management

Cloud consumption seems like it should be simple to understand, but when you get into the details, it's surprisingly complicated. My goal today is to make it easier to understand, and along the way, I'll share some insight into how the complexity of it often gets magnified.

Why Should You Care About Cloud Consumption?

From an end-user perspective, cloud consumption means either using a cloud application or consuming cloud infrastructure. The focus of this post is cloud infrastructure. Consumption in this environment is important because small inefficiencies can add up quickly when you scale. In a world that is rapidly becoming cloud first, the need to purchase capital assets and forecast usage is becoming less important. Regardless of resource location, however, understanding infrastructure resource usage and consumption remains critical because it represents an organization's ability to serve customers, innovate, and lower operational expenses.

Consumption in the Cloud

Consumption is defined as the act of "using up a resource." This concept seems simple enough when applied to cloud, but implementation differs by service providers. And understanding what exactly it means to consume a resource adds to the complexity. In the case of an individual developer who using a few services to build the latest application or of a startup that has a few employees, cloud consumption is relatively straightforward. However, at an enterprise scale, and with microservices-based architectures, the complexity of consumption grows, and the aggregate cost of cloud can significantly impact the bottom line of any IT budget.

To understand cloud consumption, it helps to understand its fundamental elements: resource types, metering, and billing.

Resource Types

Resource types are the categories by which we group IT resources. Most clouds have the following types of resources and use the calendar assumptions of 24 hours in a day and 31 days in a month for each resource. These resource types—compute, networking, and storage—are the largest drivers of cloud cost.


Compute is metered hourly, and the rate per hour is based on the value of the components that make up this offering.  

The following factors determine the value of the compute:

  • Is the compute dedicated (bare metal) or shared (a virtual machine)?
  • How many CPUs or GPUs are you using?
  • What is the speed or frequency of the compute CPU?
  • Who is the technology provider (for example, Intel, AMD, NVIDIA)?
  • What is the total memory available to your application?
  • Is local storage available?
  • What is the network bandwidth available to your compute: 10 GB, 25 GB, 100 GB?
  • What is the host operating system?
  • Are software packages and associated additional license terms needed?

All of these factors impact the per-hour rate of compute.


Network consumption is metered based on the total bandwidth consumed. All data that passes into or out of a cloud network is measured, and the associated billing is based on the amount of data transferred each month. In addition to network traffic, other associated network services influence the cost of networking services.

For networking, the value of the service is calculated based on security levels, the type of connection, and the amount of data transferred:

  • What is the type of connection to your cloud: dedicated or shared?
  • Do you have a contracted amount of bandwidth available?
  • What is the level of security on the connection (VPN, IPSec VPN)?
  • How much inbound (ingress) data is transferred?
  • How much outbound (egress) data is transferred?
  • Are you transferring data between cloud regions?
  • Are you transferring data within a cloud zone?

All of these factors impact the network cost.


Storage consumption is generally metered based on capacity consumed. All data that is stored in a cloud is metered at regular intervals (at Oracle Cloud, this is typically hourly) and is then calculated into a monthly bill. Additional factors help to determine the costs associated with storage:

  • What type of storage are you consuming: block, file, object, archive?
  • What protocol do you need to use to access the data?
  • What level of performance is required for the application (typically measured in IOPS)?
  • What level of bandwidth is required for the application (typically measured in MB/s)?
  • How fast do you need a response from the application? (Latency is typically measured in milliseconds or microseconds.)
  • Is the data encrypted? If so, at rest, in transit, or both? Have you accounted for the cost of keys or use of a key management system?

All of these factors impact the cost of storage in the cloud.


Cloud services are used over time, and the measurement or monitoring data indicates how much of a resource is used. We call this metering. Following are examples of metering on Oracle Cloud Infrastructure for each resource type based on current published pricing.

Note: Oracle provides a cost estimator to help with scenario analysis.

Compute Metering Example

One bare metal standard E2 compute shape has 512 GB of RAM, 2 x 25 GB maximum network bandwidth, and a maximum of 64 OCPUs. It is used on demand and is charged hourly, per OCPU, at a cost of US$0.03 per hour.

For a compute resource, usage is measured per CPU hourly. If a customer is using a virtual machine of a certain size or shape, the number of hours that resource was used is monitored. At the end of the month, hours of usage are calculated and multiplied by the resource rate.

Networking Metering Example

Regardless of the number of hosts, the amount of data on the virtual cloud network (VCN) is measured. Inbound data is not charged. For outbound (egress) data, the first 10 TBs are free, and over 10 TB is charged at US$0.0085 per GB transferred.

For network resources, usage is measured in GB transferred on a monthly basis.

Storage Metering Example

Ten TB of block storage is utilized and attached to a compute host. In this example, the cost of block storage is metered hourly and charged at a rate of US$0.0425 per GB month.

For storage resources, metering is based on capacity and then charged based on storage type.


Cloud billing is an accounting of all the metered usage over a given month. Although it varies, many users contribute to resource usage. Typically, organizations have a single tenant and bill for resources consumed with individual departments receiving chargebacks.

Here are some considerations for billing:

  • What type of payment do you want: pay as you go, or based on a set monthly budget?
  • How do you want to pay for cloud services: credit or invoice?
  • Who in your organization is authorized to settle services?
  • What happens if you use fewer or more resources than anticipated?
  • Are you willing to commit to a minimum amount of money or minimum usage of resources?
  • How will you identify and track who has consumed resources (users, applications, departments, projects)?
  • How do you measure efficiency? Are your servers operating 24x7? 
  • What level of usage are your users driving, and do your users power off resources when the resources are not in use?

Does It Have to Be This Complicated?

At Oracle Cloud Infrastructure, we want our customers to clearly understand the cost of our cloud services and what you can expect to pay for those services. In most cases, the resource types described here have single-dimension pricing to make it easy to understand what consumption of these services cost. We offer leading price performance with publicly defined service level agreements (SLAs) for performance, management and availability. We give you the ability to gain insight on resource expense with cost tracking tags.

Cloud consumption can be complicated, but at Oracle Cloud, we don't want billing to stand in the way of our customers' innovation. Learn more about our cloud and our simplified billing options.

Join the discussion

Comments ( 1 )
  • Adnan Akbar Wednesday, April 10, 2019
    Nice and well summarized
Please enter your name.Please provide a valid email address.Please enter a comment.CAPTCHA challenge response provided was incorrect. Please try again.Captcha