Skip to content
Notifications
Clear all

Best single sign-on for a 1000-user enterprise using AWS and Azure

4 Posts
4 Users
0 Reactions
54 Views
(@data_pipeline_rookie_43)
Honorable Member
Joined: 6 months ago
Posts: 365
Topic starter   [#27963]

Hi everyone! I've been lurking for a while, learning a ton from this community about data pipelines (my main jam is ETL with Python and Airflow). Now I’ve been pulled into a different kind of "pipeline" problem at work, and I'm hoping for some guidance.

We're a ~1000 person company, and our tech stack is heavily AWS with a growing number of Azure services. The current mess of passwords and separate logins for every tool is becoming a real operational bottleneck. Leadership wants to implement a proper single sign-on solution, and Ping Identity keeps coming up in discussions.

My team knows data, but IAM is a bit outside our usual wheelhouse. From a data engineering perspective, I'm thinking about things like:
* How well does it integrate with AWS IAM Identity Center and Azure AD for controlling access to our S3 buckets, Redshift, and Azure SQL?
* Is the setup manageable, or will it need a dedicated army of IAM specialists? We're a lean team.
* We often build internal tools (like Streamlit apps or custom APIs). How easy is it to add SSO to those?

I've read the marketing stuff, but I'm really interested in the day-to-day reality. For those of you in similar hybrid environments, what has your experience been? Any major pitfalls or "I wish I'd known" moments during implementation? Also, how does the pricing model scale for an org our size?

Thanks in advance for sharing your wisdom. I feel a bit out of my depth here, but excited to learn!

-- rookie


rookie


   
Quote
(@auditor_abby)
Reputable Member
Joined: 7 months ago
Posts: 363
 

I'm a cloud security lead at a 2k-person financial services firm. We've run a hybrid AWS/Azure environment for five years, with Okta as our production SSO for about a thousand employees and contractors.

1. **Fit for hybrid AWS/Azure enterprise**. Okta's Universal Directory and Lifecycle Management handle the sync between Azure AD (as our source of truth) and AWS IAM Identity Center reliably. Ping is equally capable here, but Okta's pre-built provisioning workflows for common AWS services were more turnkey. This setup eliminated about 80% of our manual AWS role assignments.

2. **Real total cost**. Expect $6-9/user/month for Okta's Workforce Identity + Advanced SSO package at your scale, plus a separate chunk for MFA. Ping's enterprise quote came in 15-20% higher for comparable feature sets, but included more professional services hours. The hidden cost for both is the engineering time for custom SCIM integrations beyond their template library.

3. **Setup and management load**. A basic integration with Azure AD and AWS SSO is a 2-3 week project for one engineer. Managing the system day-to-day (provisioning, de-provisioning, rule updates) takes about half a dedicated FTE's time. Neither tool is fire-and-forget; you'll need someone who understands SAML, OIDC, and your app permission models.

4. **Internal app integration reality**. For your custom Streamlit apps and APIs, both expose standard OIDC libraries. The process is straightforward: add a connector, map user groups, and configure claims. The operational burden is maintaining the user-to-group mappings in the SSO directory, not the initial code integration. Okta's API rate limits (we've hit ceilings around 10k requests per 10 minutes) can be a constraint for background job automation.

My pick is Okta, specifically if your priority is a faster time-to-value with a vast template library and your team is newer to IAM. The deciding factor would be whether you have stricter regulatory requirements that demand on-prem federation servers, which is where Ping's hybrid deployment still has an edge. Tell us if you have a hard compliance need for an on-prem component, and what your current primary directory is (full Azure AD, a mix, something else).


Where is your SOC 2?


   
ReplyQuote
(@elliek2)
Reputable Member
Joined: 4 months ago
Posts: 355
 

Thanks for this, the real cost breakdown is super helpful. When you mention "half a dedicated FTE's time" for day-to-day management, is that for your whole team to handle it, or is one person spending half their week just on SSO tasks?

I'm trying to picture what that operational load actually looks like for a team our size, because we don't have a dedicated IAM person.



   
ReplyQuote
(@cloud_migrate_tom)
Reputable Member
Joined: 8 months ago
Posts: 290
 

From a data engineer in the same boat a year ago, the day-to-day reality is that the initial setup is the real beast. It took us about three months to get Ping properly talking to both clouds and our key apps, but since then, it's mostly just adding new apps as we onboard them. We haven't needed a dedicated IAM person, but I'd say it's a solid 20% of one sysadmin's time now.

For your custom tools, if you're already in Python, adding SSO via SAML or OAuth isn't too bad. The Ping docs have decent code snippets for Flask and Django. The harder part was getting the user attributes and groups flowing correctly from Azure AD into those apps for authorization.

I'm curious, for those data services like Redshift and Azure SQL, are you planning to control access at the infrastructure level through SSO, or are you more focused on the management console logins? The approach changes the setup complexity a lot.


One step at a time


   
ReplyQuote