Chapter 7. Running Kubeflow on Azure

The Azure Cloud Platform is Microsoft’s entry into the public cloud market. Azure lets us build, deploy, and manage applications and services across a network of datacenters around the world.

Overview of the Azure Cloud Platform

We’ll start off with a review of relevant components of the Azure Cloud Platform. Most public clouds have the concept of a “region” within their datacenters, and Azure follows this pattern as well. The Azure platform defines a region as a “set of datacenters deployed within a latency-defined perimeter and connected through a dedicated regional low-latency network.” The Azure platform offers 46 regions around the world, with each region grouped into a geography.

Each geography generally contains two or more regions, and regions support distinct markets. Each market typically has its own compliance boundaries and rules for data residency (e.g., GDPR). Certain customers will be under specific data-residency rules so they need guarantees around where the data will actually live. In Figure 7-1 we can see how availability zones are arranged inside regions that are in turn arranged in data-residency boundaries.

Availability zones are one or more datacenters independent from one another, making up physically separate locations inside each Azure region. Let’s now move on to the key services offered on Azure.

Figure 7-1. Azure ...

Get Kubeflow Operations Guide now with the O’Reilly learning platform.

O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.