Reimagined firewall design, with user intent in mind, results in 500% improvement in productivity

co-authored with Rakesh Patwari, Lead Product Designer at Juniper

Juniper Security Director is a management tool that allows users to define firewall access policies for your enterprise. It also allows users to define and monitor unified threat management and intrusion prevention activities.

In our new product, Contrail Service Orchestration, that offers Network virtualization, SDWAN and Network Security, we took upon a task to super simplify how users define network security.


The new design is significantly better compared to the old model

performance improvements in the new design

Apart from the quantitative improvements in efficiency of use, the new design greatly reduces the cognitive effort for the user.


Here is how we applied design thinking and approached the problem from a user’s perspective. 
Lets consider this simple use case for this case study


Key Issues identified with the Security Director model

Three key issues were identified with the current model while achieving the simple use case above

  1. Complex multi-tiered object model: Too many objects are required to be created and used within each other. A user needs to first create an “Application firewall policy” and create “Application firewall rules” inside the policy. Then the user needs to create a “Basic firewall policy” and a “Basic firewall rule” inside it. In the rule the user needs to select advanced security and attach the “Application firewall policy” to this rule.
  2. Too many steps, clicks and a lengthy process: The UI is catered to a novice user and takes an approach of hand holding the user through the steps using wizards. This along with the need to create several objects result in too many steps and too many clicks and lot of time. The above use case requires 42 steps and 4 to 5 minutes to complete.
  3. Reflects how the device works and not aligned to the users mental model: The current model reflects how the firewall device works, which in many ways in counter intuitive to how user thinks, and results in confusing actions.
The very action where a user needs to first write an “allow” firewall rule and then attach to it an Application firewall rule with a “deny” action is counter intuitive and confusing
Complex , Multitiered object model in Security Director

This video shows the Security director UI workflow. Recorded at 2X speeds. Total time to task is almost 5 mins


The New design based on User Intent

The new design was based on these four core principles

  1. Simplify the object model:To reduce the number of objects that a user needs to create.
  2. Align to user’s mental model: The new design captures the user intent as the user would want to describe it. Internally a mapper translates the user intent to device level rules that the firewall understands
  3. Use modern interactions: like search based auto complete, drag and drop and keyboard interactions to allow faster selection of objects
  4. Reduce cognitive overload for the users: The current UI requires the user to know a lot of details about the underlying network. When the network changes the policy and rules have to be manually updated.

For the use case above, the new design completely eliminates the following cognitive overheads seen in the current model

  1. User needs to know all the below before setting out to write a rule:
  • Knowledge of the branch sites that have Finance department.
  • I.P addresses of all the firewall devices at the perimeter of these sites.
  • Manually assign the policy to each of the perimeter firewalls at these sites.
  • If there is a change in the rule or in the perimeter device, the policy has to be manually updated.

2. In the new design users do not need to assign a policy to any device. The User Intent Policy Engine refers to the latest network topology maintained by the system and intelligently identifies the firewall devices where rules need to be deployed

3. When a new site is created and has Finance department in it, the policy has to be manually updated. The User Intent Policy Engine remembers the intent and automatically deploys rules to the firewall on the new site

4. Lastly the current model requires the user to think of the right rule order where the new rule is placed. This is because the firewall stops processing the rule set when the first rule condition is matched. In the new design user need not worry about the rule order and the User Intent Policy Engine automatically places and re-orders rules to meet user intent


The video below shows the ease with which the security intent can be defined


Comparison of key metrics


Innovating through design thinking achieved significantly reduced time to complete the task, click to task and more importantly reduced a massive cognitive overhead for the user. This is what we call the “Self driving networks”