In the recent Metacoda Plug-ins 5.0 R5 release there have been a few enhancements to make it easier to sync with multiple domains (and avoid using custom code hooks):
Members of “Included Groups” are followed into other domains within the same forest.
You can opt to prefix the SAS User and Group names with the NetBIOS domain name. You might choose to do this if you have any users or groups in different domains with the same sAMAccountName and want to avoid non-unique user/group name validation errors when they get to the SAS platform.
There are more user login options available to help appropriately qualify the inbound login for the SAS user using the domain of the Active Directory user.
Some time has passed since I wrote the original post, and a few things have changed. I’m now running SAS 9.4 M3, but this post should equally apply to SAS 9.4 M2. I have also switched the Linux distribution from Debian to CentOS 7.1. I am also using a much simpler method of joining the Linux server to the AD domain, using the realmd package (previously there were lots of individual steps using the underlying packages but realmd automates most of this). In this post I’m going to outline the simpler method using realmd of course.
I’ve been spending lots of time lately on SAS® platform identity synchronizations. I’m fairly confident that I’ve done more Microsoft Active Directory (AD) to SAS Metadata Server synchronizations in the past few weeks, than I’ve done in my entire career working with SAS software! :) The reason for this is that we’ve been doing lots of testing and demos for a new Metacoda Identity Sync Plug-in we’ve built that makes it easier for people to get started synchronizing identities with SAS metadata. With all these tests and demos, the SAS metadata backup and restore facility has also been an invaluable feature for allowing us to easily rewind and repeat the process – I’ve done my fair share of backup/restores these past few weeks too :)
The idea for the Metacoda Identity Sync Plug-in came after years of writing and customizing SAS programs using the standard SAS User Import Macros (%MDU macros). I found I had built up a set of common practices I would choose from depending on the customers requirements: things like name/display-name prefixing/suffixing; tagging for deletion instead of deleting outright; login manipulation; audit reporting etc. This plug-in is a way of packaging those practices up, as configurable options, with both a point-and-click and a batch interface. The outcome is an ability to rapidly implement identity synchronization, for a new or existing SAS platform installation, in a matter of minutes (rather than hours or days of writing code).
It has been a very rewarding experience building this new plug-in, and the feedback we’ve had so far has been very positive. Some of the interesting challenges along the way included:
Making it easy to get started, but also flexible enough to handle some of the more specific requirements we see with our customers. The point and click interface includes the common options, but we also added support for customers to tweak things by dropping their own SAS code in at key points in the process too.
Letting people interactively visualize and review the changes before they are made, adding and removing exceptions as required, and building a configuration that can be used in batch processes too.
Working within AD resource limits whilst extracting reasonably large subsets of identities for synchronization with SAS. Some of our tests included pulling out many thousands of users (40K+), including groups that contained several thousand users each.
Providing support for encrypted connections to AD via LDAPS, or LDAP with STARTTLS.
Generating audit reports of the process, so you can track what changes occurred when, and with all of the information that led to those changes.
Writing our first commercial plug-in that updates metadata (all our other commercial plug-ins to-date have been read-only). In this plug-in we have opted to only update metadata via the standard, unmodified, well known and trusted SAS %MDU macros. Whilst we have lot of experience with the SAS metadata model, we decided to give our customers a gentle introduction to Metacoda driven metadata updates.
If you’d like to see the Metacoda Identity Sync Plug-in in action, here’s a short 10 minute screencast. I show the initial configuration, building an Identity Sync Profile, and a small initial load of AD users, driven by the selection of an initial set of AD groups. That saved profile can then be re-used for further interactive synchronizations (adding, updating and deleting identities as appropriate), as well as being used to drive regular batch synchronizations (topics for future screencasts perhaps?).
We’ve been getting some great feedback from the people we have shown so far, so I hope you’ve found this video interesting too. If you’d like to find out more about this new plug-in, or any of our other Metacoda Plug-ins, please contact me, or visit the metacoda.com web site. We’re still taking on beta testers for the the upcoming Metacoda Plug-ins 5.0 release too.
Continuing on the theme of configuring a SAS 9.4 M2 platform on Linux to use Integrated Windows Authentication (IWA), in this post I’m going to jot down some notes on steps 12-15 – configuring the SAS mid-tier on a Linux server for IWA with fallback to form-based authentication (when IWA is not available). This includes delegation, so that IWA users of mid-tier apps like SAS Studio are able to get IWA access to a SAS Workspace Server (and avoid having to store their passwords in metadata or switch to using SAS Token Authentication).
If you’re wondering what happened to steps 1-11, I’ll try get to those earlier steps in future posts. I’m starting at step 12 because someone recently asked me a question about configuring an IWA mid-tier and so it seemed like a good idea to get this blog post done first. Of course, when actual implementing, it’s always good to start at the beginning, building up the foundations, and verifying those first steps are working well before moving on to the next steps. So these steps assume you already have a working implementation where SAS desktop applications (like SAS Management Console & SAS Enterprise Guide) are able to connect to the SAS metadata server using IWA, and also to get IWA access to an appropriately configured SAS Workspace Server.
I have found the best mid-tier related documentation resources for this type of configuration are these ones:
One of the reasons I’m writing this post is to get down some notes on a config that worked for me. The documents referenced above cover a variety of scenarios including plain basic web authentication with an XML file-based UserDatabaseRealm, an LDAP JNDIRealm, IWA (SPNEGO) without fallback, as well as fallback to form-based SAS authentication. Getting the right mix of settings, that didn’t conflict with each other, took me a long time to determine (my mid-tier takes about 20 minutes to restart whenever I want to test a modified config). Along the way I encountered pop-up basic web authentication dialogs when IWA should have worked, and infinite browser-refresh loops for the SAS Logon Manager when IWA was disabled in the browser and I was expecting fallback to SAS authentication. This post is about the final config that worked for me. I know I’ll be referring to this post again, and I hope it proves helpful to others too.
One of the other reasons I built this configuration was to find out what happened with SAS Visual Analytics Guest Access in an IWA fallback configuration like this. Essentially, I wanted to find out if I could get maximum flexibility by supporting IWA users, form-based authentication users, and guest/anonymous access all at the same time.
One of the reasons I wanted to test this was a reference I remembered seeing in the SAS documentation. The Web Authentication section of the SAS 9.4 Intelligence Platform: Security Administration Guide, Second Edition, lists one of the limits of Web Authentication as “Not compatible with anonymous access”. This is also repeated in the PUBLIC Access and Anonymous Access section too.
It makes sense that anonymous access is not compatible with web authentication in a standard non-fallback configuration. If authentication is automatic and it fails then access is denied. An IWA fallback configuration is slightly different though – you have a choice whether to do web authentication or SAS authentication (e.g. IWA or non-IWA). If you choose SAS authentication then perhaps anonymous access might still be available as an option. I decided to test it out.