Works well for small businesses but is not stable enough for a company of our scaleWe have been trying not to use the solution. It is used for a specific use case, which is around authenticating M365, and we are trying to see if we can get out of using it, but that is only because our environment is extremely complicated. Entra ID is not battle-tested or stable enough to support a business of our size. There are some design issues specifically around support for legacy services. We used to be part of Microsoft, so we have about 15-year-old services sitting in our data center that still need to use legacy LDAP authentication. The way we currently have the environment set up is for one very specific domain. I am using a domain for specific context here to keep it simple. We have 36 Active Directory domains, and that does not include the child. We follow the least privileged access model. Our environment currently consists of using AD Connect to synchronize objects from our corporate tenant into Entra ID, and then from Entra ID, we wanted to stand up Azure domain services as a possibility for retiring legacy LDAP services. The issue with Entra ID specifically is that the way it replicates objects out of its database into the Azure domain services Active Directory tenant or Active Directory service is that it uses the display name. This is a bad practice, and it has been known as a bad practice even by Microsoft over the past decade, so the design is not good. The issue with replicating based on the display name is that when you are coming from an environment that uses a least privilege access model, where you want to obfuscate the type of security account being used by hiding it behind a generic display name, instead of myusername_da, myusername_ao, etcetera, to have an idea of what accounts are being used when they are logging in, it is unable to reconcile that object when it creates a new domain. If they all have the same DM, you end up with quadruplicates of each user identity that was replicated to it from the directory. Those quadruplicates or their same account names, as well as the display names within the cloud domain services directory, have a unique identifier with the original account name attached. What that does is that it not only breaks that LDAP legacy authentication, but it also drives up the cost for your customers because you are paying for each additional seat, additional user objects that are created, or additional users. You also cannot tell any of those accounts apart unless you dive deep into the user object to peel back what type of account that is to map it back to what came from on-prem itself, so the service is completely useless. What we have done in our case is that we do not really need Entra ID. We have Okta, so we use an Okta LDAP endpoint. That does exactly what we need in using SCIM, which is the technology that is able to take identities from multiple dynamic providers and merge them together into a single record. It is able to act as an official LDAP endpoint for the business, so legacy apps work. We do not have a problem. Microsoft could learn from that. Entra should allow for external MFA providers rather than forcing you into a walled garden and the Microsoft ecosystem. Flexibility is a big thing, especially for companies of our size. A big issue for us is that we want the identity to be in Entra for sure, but we want it to come from Okta. We want the authentication and stuff to work, but we want Okta to control the PIM rules. We want it to do the MFA and all those things, but Entra does not play nice with others. Okta has engineered some ways to get it done, but it is not as full-featured as we would like it to be. Microsoft should do what they do with some other partners such as Nerdio and Jamf where they have their own version of a service, but they are still partnering with those other companies to at least add options on the market. Fully customizable UARs and Azure Secure Identity Workflows would be great. Currently, you can do it if you cobble together a bunch of Azure functions and use Sentinel. If you are sending logs to Sentinel and are able to match patterns and run automation based on that, it would be great. They can help with a solution that abstracts away a lot of that complexity across multiple services into exactly what IIQ does. I could definitely foresee Entra being the choice for identity for pretty much all cloud providers if they can focus on the areas that SailPoint's IIQ does. A big pain point for a business of our size by being in Okta is that we do not have the same workflows that we have between IIQ and AD. With the amount of data that our company generates, we wanted Sentinel. I had their security department onboard, and it was going to be millions a month just to use Sentinel, but we could not use it, so we decided to leverage Splunk and a few other SIEM providers. They should also stop changing the name of the product.
