Azure resources and services can be accessed using different mechanism. From LIVE accounts and/or AD users to access tokens (Shared Access Signatures (SAS) and Shared Access Policies).
In this way, if the security of that component is compromised, we can invalidate that token, without affecting the rest of the application. Of course, managing all this tokens can be a nightmare, similar with the one that we have when we need to manage users from Active Directories (ADs).
Token based authentication and authorization provides delegated access to any service or resource that is exposed by Azure. It helps us a lot when we need to provide access to external systems. We have a granular and fine control to all resources that can be accessed in this way. Each token can be generated on the fly, based on our needs.
You don't want to share the 'full admin' rights to all external systems or components.
From security perspective, this is a very useful thing. We don't need to create users, manage password, registration form and maintain the user list.
It is very tempting to use such a solution inside the system. Each internal component that access a Azure resources will be able to access it only based on a unique token that was generated only for that scope.
For example, if two components needs to access a specific Service Bus Topic, than for each components a unique token will be generated that allow them to access that specific resource.
In this way, if the security of that component is compromised, we can invalidate that token, without affecting the rest of the application. Of course, managing all this tokens can be a nightmare, similar with the one that we have when we need to manage users from Active Directories (ADs).
To be able to manage a system only based on tokens, we need to be sure from the 1st moment that we have a powerful configuration and management system. Otherwise, configuration and deployment can become a nightmare.
Why token based authentication and authorization is so powerful?
Nowadays, applications and system are based on smaller and smaller components. Each components can manage only one functionality. In an ecosystem where you have 50-100 components and subsystem you need a powerful and flexible solution. This solution needs to allow you to manage the security without blocking the entire system in a case of a security breach.
Why to not use the 'root' access keys?
For small systems we could use the root keys internally without any kind of issues. The problems appears if you need a more granular access each resources, based on the component that needs to access it. On top of this, if you are working in a bank or life science industry, than the rules are very strict. In such industries you need to have a granular access to each resource.
Also, if the security of a component is compromised and you are using the 'root' keys, than you will have a short period of time when the system will not be available. You will need some time to regenerate the unique keys (tokens) and distribute them to each components. Yes, it is true that for 'root' access keys you have a primary and a secondary one, but when a component is compromised that contains both keys, than you need to reset both of them.
Comments
Post a Comment