IaaS 101: Types Of Compute

IaaS Design
7 min readApr 30, 2021

An introduction to IaaS Compute at IBM Cloud

Compute is a vast, intricate and critical piece to our Cloud platform. In its simplest terms, compute (or computational power) is the process of how computers take tasks (jobs), that are made up of data and complete (or compute) them in a sequence.

As simple as it sounds, this ability helped us land on the moon and enables our modern world today. Our platform, along with others like AWS, Microsoft Azure and Google Cloud, offers customers various levels of compute services (and a plethora of other products) based on their companies’ needs and budget; meaning that customers can off-load the substantial costs of purchasing, running, securing and maintaining the compute infrastructure needed to run their applications and businesses.

This is made possible because of data centers. These buildings, located in various regions all around the world, house rows and rows of servers (groups of computers) and are connected together through switches and routers (networking). They have access to redundant cooling, redundant power, an internet connection and skilled admins and techs who keep everything going. Customers can rent as much or as little as they need and focus on building their businesses, not maintaining infrastructure.

As more companies move to the cloud, common customer use-cases are:

  • Building cloud-native applications (specifically designed to run consistently across private, public and hybrid clouds), or extending legacy applications into the cloud space.
  • Expanding capacity (due to limited on-premise hardware availability) or taking advantage of the rapid scaling capabilities of cloud infrastructure.
  • Easier access to (and consumption of) cloud services, products, and features, without having to rely on complicated high-touch workarounds.
  • Modernization of legacy monolithic applications, to take advantage of the latest and greatest infrastructure or containerization technology.
  • Additional resiliency measures, including data security and disaster recovery.
  • And a very important one… price. Cloud computing is a more affordable option when compared to on-premise infrastructure.

Compute at IBM Cloud is broken up into different specialized teams. Our specific IaaS (Infrastructure as a service) compute design team works on: VPC, VMware, Virtual Servers, Bare Metal Servers, and recently Networking appliances and interconnectivity (which enables everything to be connected together). The other teams that round out the compute catalog are: Power VS (Power), IBM Z (Hyper Protect Virtual Servers), Kubernetes (IKS), Red Hat OpenShift (ROKS) and finally Code Engine, which is a server-less compute and is its own specialty not covered here yet.

At the heart of our compute products are two foundational technologies on which everything builds — Classic infrastructure and VPC (Virtual private cloud) infrastructure. Classic infrastructure (Bare metal for classic) offers many levels of customization and is commonly used for moving legacy applications into the cloud as well as extending existing on-premise applications and high-power cloud computing (gaming, finance, and databases). VPC infrastructure (Virtual servers for classic, Virtual servers for VPC) is the modern solution for multi-zone cloud native applications, regulated & highly secure industries (due to the additional security/privacy measures), as well as modernized hybrid/multi-cloud infrastructure.

Bare Metal for Classic

A physical server (computer) that is entirely dedicated to its customer. These devices are highly secure and stable, but also generally the most expensive option. Due to the physical nature and availability of components in different locations they often require longer lengths of time to set up and take down.

Virtual Servers for Classic

A bare metal classic server that is virtualized and divided (via a hypervisor). This allows you to “rent” portions of, or an entire physical machine, making them a good option for volatile workloads (due to speed of provisioning) or low-intensity compute applications.

Virtual Servers for VPC

By adding a hypervisor to a Bare Metal Server it creates a VSI or Virtual Server. While VSI for VPC is not as mature as its Classic counterpart, users can benefit from faster provisioning times, improved performance, greater control over their environment, enhanced security and privacy, and reduced interference from noisy neighbors.

Bonus: VPC Networking

VPC (virtual private cloud) infrastructure allows you to create your own logically isolated public cloud wrapped in a private secure networking environment. This is the modern approach to networking for cloud infrastructure, and allows precise control over network traffic in and out of your cloud environment.

Once a customer decides on a foundational technology, there are two additional platforms that a customer can choose from: VMware which is the older and more familiar technology, and containers, which run on virtual machines but are a modern approach for running cloud native applications (Kubernetes and OpenShift).


An external company that has ongoing partnerships with each major cloud provider, helping their customers take existing on-premise VMware runtimes and move them to the cloud of their choice with access to modern services. VMware today runs on Bare Metal Classic using a special hypervisor called an ESXi.

Analogy — this product would be considered a house, or a condo. With access to both dedicated and shared environments, customers have many customizable options, but is more expensive than the container based products and usually only a sensible option for customers with existing VMware runtimes.


IBM Cloud Kubernetes Service (IKS) enables users to run managed Kubernetes clusters using a variety of underlying hardware types. Kubernetes is an open source container orchestration technology that simplifies the use and management of many containers. Kubernetes automates deployment, scaling, healing, and management of containerized applications.

Analogy — this product would be considered a motel to start (but if users are willing to invest time and materials it can be built up to a fancy hotel). It is extremely cost effective, and a no-frills approach to running containerized applications with a robust ecosystem of open source tools you can add to it.


Red Hat OpenShift on IBM Cloud (ROKS) is a Kubernetes based container platform that enables users to run managed OpenShift clusters. This product allows developers to develop, deploy, and manage applications across hybrid and multi-cloud environments.

Analogy — this product would be considered a 5-star hotel. It builds as a user experience on top of Kubernetes and enables developers to harness the power of Kubernetes products.

Bonus: Dedicated & Shared
A dedicated environment is an entire server or host where the customer has full control over its resources and how they are utilized. They are high performance and extremely secure, but come at a premium cost in comparison to shared.

A shared environment allows a customer to rent a piece (small or large) of a server or host that is shared with other customers. They are very cost effective, and a good option for many customers depending on the types of applications they need to run in the cloud.

Analogy — dedicated would be considered a house. It is isolated from its neighbors (does not share a wall) and the owner has full control over what they would like to do with the house. Shared would be considered a condo, hotel, or apartment. The tenants share resources, are close together in space but is also cheaper than a house.

Another important piece of our compute catalog are the Power and Hyper-protect products which are for special use cases such as: artificial intelligence, applications with highly sensitive data (confidential compute), high-power compute, and extending on-premise Power workloads.

POWER systems Virtual Servers

Power Systems is a family of servers from IBM that are based on the POWER processor architecture and are specifically designed for the most data-intensive workloads that surpass the capabilities of x86 architecture. Previously, Power servers have been predominantly used on-premise, but today IBM Cloud brings this enterprise-grade platform to hybrid and multi-cloud deployments.

Hyper-protect Virtual Servers

These are x86 servers that provide a confidential environment to protect your data. They are designed to address the top-most security concerns, and not even IBM Cloud administrators have access once the resource is provisioned.

Wrapping this up
Yes, compute can be daunting, expansive, and complex — remember you are not on your own. Our team works together to understand, create, and improve our customer experiences. Hopefully this article has helped give a small introduction to the world of compute at IBM Cloud.

– Written by: Austin Edwards, Jenny Lanier and Josef Bodine
– Illustrations by: Austin Edwards



IaaS Design

The IaaS Design team is a specialized group of designers and researchers within IBM Cloud.