Amazon AWS ALB – Single login authentication

Amazon AWS ALB – Single login authentication

Amazon AWS ALB
Atlassian customers hosting their on premise applications on AWS can now enable authentication with their AWS ALB without a second login step.

Table of Contents

Here is some context in case you’ve landed here from a Google search. In March 2021, we launched a new family of Atlassian plugins: AWS ALB and Amazon Cognito Authentication for Jira and Confluence. For shorter, we will call this app AWS ALB Authentication.

Since December 2021, all versions of the AWS ALB Authentication apps have been approved for Data Center and can be used with full guarantees by customers hosting their applications in multi-node on premise environments.

AWS ALB and Amazon Cognito Authentication for Jira

With ALB Authenticate, any organization already hosting its Atlassian applications in AWS can use their Amazon infrastructure to create a Single Sign-On flow.

The advantage? A much more secure infrastructure. Your Jira or Confluence will only be hit by traffic that has already authenticated to the AWS Application Load Balancer.

In general terms, the approach resembles very closely what can be obtained with a more traditional SAML SSO setup. But let’s look at the differences.

The role of AWS Application Load Balancers in your Atlassian Stack

Before diving into the hardcore authentication stuff, let’s do a quick reminder on the role of Application Load Balancers for enterprise customers of AWS and in the Atlassian stack specifically.

The major role: distributing user traffic among available nodes

An Application Load Balancer and its role in an Atlassian Data Center architecture

In Atlassian Data Center applications, a load balancer is placed to distribute users between the different nodes where the instance is installed. The load balancer routes traffic so that it never becomes a problem, usually relying on health checks and server diagnostics. This ensures that the application traffic doesn’t become an excessive stress, which typically happens during peaks of concurrent users, or when one or more of the nodes are down.

When Data Center applications are hosted in AWS, they are typically protected with an Application Load Balancer (ALB). One worth mentioning is that they grant sticky HTTP sessions for a continuous user experience.

Using AWS Application Load Balancers and Amazon Cognito in your Jira authentication flow

Application Load Balancers can also be configured to authenticate users, usually in combination with Amazon Cognito identity pools or with cloud Identity Providers that support the OIDC protocol, such as Azure AD, Okta or GSuite.

Read the AWS documentation on How to authenticate users with an Application Load Balancer

Main advantages of AWS ALB Authentication

Here are the main reasons why you should consider this setup if you’re already hosting your Atlassian applications in AWS.

  • An exceptional security layer. Use your Application Load Balancer as a firewall that protects your Atlassian applications from unauthenticated traffic.
  • Create a seamless flow for your users. With the resolution app, users don’t have to authenticate a second time when they are redirected to the Atlassian application.
  • Provision and deprovision users! Since its second version, the AWS ALB app ships with a User Synchronization module. This allows to create connectors that synchronize your local Jira or Confluence user directory with your cloud user directory. Your users will always be up to date and your admins will never have to go into Jira to create a new account or deactivate an existing one!
  • Integrate with your existing cloud infrastructure. AWS ALB Authenthication fits perfectly if your company is already using OpenId Connect, Amazon Cognito, or any OIDC compliant Identity Provider.
  • Easy configuration. Select which token and token claim should be used to authenthicate into Jira or Confluence. With predefined templates for Amazon Cognito, Okta, Azure, and GSuite!

But let’s have a closer look at how the app fits into the picture.

Basic flow with AWS ALB Authentication

Diagram flow for an authentication using an AWS ALB and the resolution app
  1. An unauthenticated user tries to access “jira.awesomedomain.com”, but doesn’t reach Jira.
  2. The AWS Application Load Balancer intercepts the incoming http request before it reaches the Atlassian application and redirects the user to the Identity Provider for authentication.
  3. After a successful authentication, the user is redirected back to the ALB along with an ID token which allows the ALB to obtain more user details. Finally, the user is redirected to your Atlassian product along with the extracted information from the ID token.
  4. Resolution’s app uses the information passed along by the load balancer to log in the user. No additional login is required. That’s a good job, right?

Comparison with a SAML SSO authentication flow

Flow diagram for an SSO authentication into Atlassian applications

Let’s now look at a SAML SSO flow from a very high level to understand the main differences between the alb authentication and a more traditional SSO solution.

  1. An unauthenticated user tries to access “jira.awesomedomain.com”, and hits the Atlassian login page
  2. The SAML SSO app redirects the user to the IdP login page
  3. The user authenticates on the IdP
  4. If the authentication is successful, the user is redirected back to Jira and directly logged in.

As you can see, the main difference is that in a SAML SSO authentication flow an unauthenticated user will always arrive at the Atlassian product in the first place. This has some implications:

  • You have to trust the security of your Atlassian product. If there’s a loophole allowing unauthenticated access to the instance, an unauthenticated user could exploit it.
  • Public pages in the Atlassian product will be accessible to anybody. And Jira has many of them (if you haven’t configured your SSO app to protect them).
  • Higher traffic load on the Atlassian application. All unauthenticated traffic hits Jira or Confluence directly before the logging event. Instead, the Application Load Balancer intercepts the incoming traffic and only lets in legitimate users. In scenarios where scripts or attackers try to ddos your Atlassian application, they will “only” be able to hit the load balancer. Your Atlassian product will be left alone.

How to configure your AWS ALB to handle your Atlassian users

Let’s see how to configure the app to handle users. We’ll first look at how to set up user authentication. Next, we’ll have a look at the different options for user provisioning and deprovisioning.

In this guide, we’re assuming that you have successfully hosted your Atlassian Server or Data Center applications in an AWS server and that you have an ALB or ELB in your Amazon infrastructure.

Setting up the Amazon ALB authentication process in three steps

Configuring the connection of the ALB with your IdP is quite simple. You will always need to follow the same three steps, although the details will change a bit from vendor to vendor.

Here’s a list of the specific guides:

Step 1. Select your IdP

It all starts with selecting the Identity Provider where your user directory is stored. If you have something other than Cognito, Azure, Okta or GSuite, just select Other.

IdP selection page Amazon Cognito

Step 2. Configure how the tokens will be processed.

In this step, you need to tell the app where to look in the JWT that the ALB will exchange whenever a user tries to login. The screenshot shows the most common setup.

JWT token processing settings
  • Token Header Name: What’s the name of the token header that contains the user info?
    • Cognito: x-amzn-oidc-data
    • Azure: x-amzn-oidc-accesstoken
    • GSuite: x-amzn-oidc-data
    • Okta: x-amzn-oidc-data
  • Username Claim: Which field contains the username?
    • Cognito: email
    • Azure: upn
    • GSuite: email
    • Okta: email

Step 3. Security checks

Configuring security checks Amazon ALB authentication

To ensure that the JWTs are not forged, you can have the app check the identity of the issuing IdP and your specific ALB.

  • To check the URL of the issuer, access your IdP as instructed in the corresponding guide (check the link at the start of this section)
  • To check the Amazon Resource Name (ARN), go to the Description tab of your Load Balancer in AWS.
where to find the amazon resource name (ARN)

Configuring user provisioning

There are two patterns for provisioning users with our AWS ALB Authentication app:

  • Syncing the entire user directory, without authentication
  • For individual users, during login

The two methods can perfectly coexist. For example, you can decide to update users as they login to grant them correct permissions to their Jira projects and Confluence spaces. But you can also synchronize the directory every night to ensure that you don’t have discrepancies in who should have an Atlassian account.

When in doubt, remember the major differences:

  • Individual user updates during login can’t remove usersdirectory syncs can.
  • directory syncs will also update users who don’t have the habit of going into Jira or Confluence. This can be useful if someone needs to mention them or assign them work on a new project before they login.

Synchronizing the entire user directory

In this case, you will build an IdP connector to sync the entire directory.

Don’t be scared! This doesn’t mean that everyone in the company must get an account. You can filter by groups, and the most usual way is to only provision users who are assigned to Jira/Confluence in the IdP.

The steps for building the connector vary a lot from IdP to IdP, so we’ll just point you to the specific guides:

Also, bear in mind that these connectors are needed for the individual update using the REST API connection.

Once you’ve setup the connection, you can decide whether you want to:

  • trigger the syncs manually from the configuration UI
  • or scheduled the sync to happen at regular intervals. For example, every Saturday night.

Create & update individual users during authentication

If you choose to update users as they login, you can still choose between two different methods in the configuration.

  • Use the information sent in the claims of the Amazon Load Balancer (Just in Time Connector)
  • Connect to the IdP via REST API (IdP Connector)

The first method can be enough for simple needs, and is easier to setup. The second will satisfy even the most demanding use cases with dozens of user attributes.

In both cases, you need to enable User Provisioning with User Sync. To do that, scroll to User Provisioning Settings and enable the toggle as shown below:

Enabling User Provisioning with User Sync
Update During Login with a Just in Time Connector
  • Move over to User Sync and click on Create Connector. Then, select Just-in-Time from the dropdown.
creating a Just in Time connector
  • In most cases, you want to create a new empty directory with the button
Where to find the "create new empty directory" button
  • Next, let’s move to the Provisioning Settings tab and map the claims sent by the Amazon Load Balancer with the Atlassian user. At least, map the following:
    • Username
    • Full Name
    • E-Mail Address
  • After saving the configuration, you only have to select the Just in Time Connector in the dropdown that shows under User Provisioning Settings.
Select the Just in Time connector
  • Finally, save the configuration again.

For more detail, here’s the step-by-step guide:

Just-In-Time Provisioning – AWS ALB & Amazon Cognito Authentication

Update During Login with an IdP Connector

Using an existing connector for your IdP, you can also activate a lookup function that triggers whenever a user tries to login.

What you have to do is instruct the connector to look the user up with a specific attribute:

  • For Azure AD: upn
  • For Okta, Google, Cognito and Keycloak: email

Now follow these steps In the User Provisioning Settings of the AWS ALB & Amazon Cognito Authentication app:

  • Select the connector you want to choose
  • Edit the operation “Attribute as received from the AWS ALB”
  • Replace the Source Attribute Name with the field indicated above, and apply the changes.

Here’s the step-by-step guide: User Provisioning with UserSync – AWS ALB & Amazon Cognito Authentication

Conclusion

If you have successfully migrated your Atlassian instances to AWS, then AWS ALB Authentication is the easiest way to prioritize security and performance while maintaining a frictionless user authentication experience.

Have a try now for free for 30 days, and get in touch with our support team if you need any help figuring out the exact details of your implementation!

Subscribe to our newsletter:

Related articles: