Responding to the Market
Goal: Create a Ring Deployment (automated patch testing and deployment) system better than Microsoft's version.
My Roles: Product designer, researcher, strategist
Impact: Many major sales contingent on this feature.
Ring Deployment for Ivanti Neurons
Ring Deployment – allowing users to “test” patch content like any other software, in 2-3 successively larger “rings”,- was the biggest customer ask, once we rolled out Patch for Neurons itself.
We wanted to improve on what Microsoft did. I think we achieved it.
Situation
Once Patch for Neurons got out into the market, the biggest single “ask” was Ring Deployment (sometimes called Rollout or Campaign deployment) – allowing the users to structure the rollout of patches in “Rings”
-
A small “Test” ring of machines where the user would make sure there were no catastrophic problems with the patches
-
A larger, optional “early adopter” ring where the user would make sure the content was compatible with the full network
-
“Production” – the entire network.
Task
Add a “ring deployment" feature to Patch for Neurons.
f
Make it better than Microsoft’s ring deployment feature, which arbitrarily divides a network – whether 100 machines or 100,000 – into three groups (1% into a Test group, 9% into “early adopters”, and the remaining 90% into “Production”.
Actions
There were two real functional requirements:
-
Allow the user to view how rings were performing, and see problems emerging in a testing process
-
Divide the network’s machines into “rings”, of roughly the same proportion as MIcrosoft’s model – but avoid some of the problems that model causes.
Ring Operations View
The Ring Operations View allows the user to see the progress of the patch content as it moves through the rings (figure 1), and the state of the devices who are the parts of the rings
One of our design iterations. It was not a successful one – it was no advance over the on-premises version. It took minutes to design, andseconds to dispose of. (Pardon the image quality – after five years, it’s a screenshot of a screenshot).

T
The Ring Operations View Patch State shows the status of content as it moved through the deployment process. It enables the user to see evolving problems in deployments of thousands, even tens of thousands, of individual items of content.
This is one of six primary pages in Patch for Neurons.

The “Device” state of the ring operations view and monitor the status of attempted installations, across the network and (via the tabs) within each ring.
The active graphics allowed the user to zoom in and redraw the grid to highlight the devices that were having problems – the ones they were actually interested in.
Ring Config List
As in most companies’ lines of software products, the word “configuration” gets overused.
For simplicity’s sake, I opted to have every patch settings configuration have a 1:1 relationship with a Ring Configuration. Every Patch Configuration would show up in the Ring Config list – but if it didn’t ‘have rings configured, it’s should show as “Unconfigured”. Otherwise, the top level ring operations state would display.

Ring Assignment Wizard
Before rings can operate, they have to be configured.
Microsoft’s automatic ring configurator splits the user’s environment more or less arbitrarily into three rings:
-
A Test ring (1% of the endpoints)
-
An Early Adopter ring (9% of endpoints)
-
The Production ring, with the remaining 90%.
This produces some problems:
-
There are some users and machines – C-suite execs, salespeople making critical presentations, vital servers – that the user absolutely does not want included in a test ring.
-
Microsoft’s model selects machines at random – without explicit regard for the applications on those machines that are being patched in the first place.
So I designed a multi-step ring configurator – a “Wizard”. It walked the user through five steps:

Step 1 - Specify the percentage of devices the user wants in each ring. This can be skipped. The user also specifies the minimum number of instances of each product (e.g. Chrome, Office 365, OSX) to be tested in each ring to create a valid test sample.

Step 2 - Specify Devices and Device Groups that are to be included in the Test and Early Adopter rings. If Step 1 is skipped, the user explicitly adds groups or devices to the rings.
If the user specified random sets in Step 1, these assignments will be in addition to the random set.

Step 3 - Specify device groups or devices to exclude from rings.

Step 4 - Ensure that, after making adjustments in steps 2-3, there are still enough instances of products being tested in each ring.

Step 5 - Show the user a summary, and alliow the user to adjust any device assignments.
Results
Complex as it appears, ring deployment has performed well in usability tests. It should be rolled out in April 2025.