- Hands-On Artificial Intelligence for Beginners
- Patrick D. Smith
- 263字
- 2025-02-23 08:58:39
Cloud computing essentials
Often, on-premise GPU clusters are not always available or practical. More often than not, many businesses are migrating their AI applications to the cloud, utilizing popular cloud provider services such as Amazon Web Services (AWS) or the Google Cloud Platform (GCP). When we talk about the cloud, we are really talking about database and compute resources, offered as a service. Cloud solution providers such as AWS and GCP have data centers across the world that store data and run computing jobs for people remotely. When your data is in the cloud, or when you are running a program in the cloud, you are really running or storing in one of these data centers. In cloud terminology, we call these data centers or cluster of data centers regions.
Cloud services are divided into three different offering structures:
- Infrastructure as a Service (IaaS): Raw computing and network resources that you can use to build infrastructure, just as you would locally
- Platform as a Service (PaaS): Managed services that obfuscate away infrastructure components
- Software as a Service (SaaS): Fully managed solutions, such as online email
In this section, we'll cover both IaaS solutions, as well as PaaS solutions. While cloud providers do offer SaaS solutions for AI, they are a bit too high level for our needs. In this section, we'll discuss the basic tools that you'll need to utilize the compute power of the cloud. Towards the end of this chapter, we'll discuss cloud computing in more detail in the Maintaining AI applications section.