Hey everyone. New to Perimeter 81 here, still evaluating it for our team.
We're setting up the always-on VPN, which seems great for security. But I need a way to exclude a few specific corporate laptops from automatically connecting. These are used for testing local network configurations and some legacy on-prem tools that break with the VPN tunnel.
Is there a way to create an exception policy based on the device itself, not just a user or group? Or do we have to turn off the always-on client completely on those machines? Looking for a clean setup without manual toggling.
You can absolutely create device-based exceptions in Perimeter 81, and you're right to look for that instead of toggling the client on and off. The clean way is to use device posture checks.
Go to the policy section and look for the conditions tied to "Always On." You can set a rule where the VPN won't initiate if a specific file exists on the device, or if a particular registry key has a certain value. You'd push that file or key to your test laptops via your regular device management tool. That keeps the client installed and running, but it won't trigger the tunnel.
Just be aware that this puts those devices outside your normal secure perimeter, so you'll want to make sure your other security controls on them are tight.
Keep it civil, keep it real
That device posture trick is clever, but I'd be careful with it. You're creating a permanent bypass condition on those laptops.
Better approach: set up a dedicated network for those devices in Perimeter 81 that doesn't have the always-on rule attached. Move the test laptops into that network via their device ID. That way they're still under management, you get logging, and you can flip them back to a secure network later without redeploying files or registry keys.
garbage in, garbage out
I agree that the dedicated network approach is more manageable for a defined set of devices, especially if you treat these test laptops as a distinct asset class. However, it's important to clarify that the device posture method isn't necessarily a *permanent* bypass. You can design the check to be dynamic.
For example, you could set the condition to look for a registry key value that your device management script updates daily. If the key is set to "TEST_MODE=1", the VPN stays off. Once testing is complete, your script flips it to "TEST_MODE=0" and the next posture check will enforce the always-on policy. This gives you programmatic control without moving devices between networks, which can be useful if your network structure is complex.
The network-based method is cleaner for straightforward cases, but the posture check can be more granular if you need the exception to be conditional or time-bound. Both are valid, but your choice depends on whether the exclusion state should be a property of the device or a property of its assigned network segment.
Your data is only as good as your pipeline.
That's a solid point about dynamic posture checks. The scripted registry key toggle is clever for temporary exclusions, and it's something I hadn't considered.
But it makes me wonder about the operational overhead. Now you're managing a script, its deployment, and a schedule, versus just moving a static device to a different network segment. For a handful of test laptops, maybe the script is fine. But if this logic ever needed to scale to other use cases, like excluding devices during specific maintenance windows, I'd lean toward the network method for its simplicity in the admin console.
Either way, you've highlighted the core decision: is the exclusion state a device attribute or a network assignment? Great way to frame it.
Data is the new oil - but it's usually crude.
This is exactly the problem I'm hoping to avoid with our own VPN rollout. That clean setup without manual toggling is key.
The device posture method user946 mentioned seems promising, but the dedicated network idea from user351 feels like it might be easier to audit later. If you move a laptop to a "Testing" network, you've got a clear record in the Perimeter 81 logs of *what* you did and *when*. A registry key buried on the local machine feels a bit invisible from the admin side.
Have you considered if you'll need to switch these laptops back and forth frequently, or is this a one-time, permanent assignment for their specific role? That might decide which method is less hassle.
null
The dedicated network approach isn't the clean break from overhead you think it is. You're just trading file management for network object management.
Now you have a special network sitting in your config forever, accruing policy baggage. What happens when someone inevitably attaches a new access rule to that network six months from now, forgetting its purpose? You've created a policy anomaly that's harder to audit than a registry key, because the *intent* is buried in a ticket, not in the device's own state.
Logging is a fair point, but that's a feature of any decent MDM - you'd log the registry key change, too. The real failure mode is that networks become political, not technical, constructs.
Your k8s cluster is 40% idle.
You're right about networks turning into policy dumping grounds. I've seen it happen where someone needs to block a SaaS app and just picks the "Test_Devices" network because the name looks harmless. Suddenly your test laptops can't reach Jira and you're debugging for hours.
But the registry key method isn't a policy of record either. It's a side channel. At least a network assignment is a declarative state inside the same system that enforces the rule. Your MDM logs about the key change live in a completely different silo. That's two systems to query during an incident, not one.
Both are hacks. The real problem is that "always-on" shouldn't mean "no escape hatch." The tool lacks a first-class bypass toggle with proper approval workflows.
Least privilege is not a suggestion.
Agreed, it's a flaw in the product's design. We ran into this with another vendor and ended up adding a cheap "always-on exception" SKU to our contract, priced per device. It forced them to build a proper toggle in the UI because they had to support the billed feature.
Suddenly, we had an audit trail *and* the cost made us think twice before adding exclusions.
Forcing a vendor's hand with a line-item on the contract is a brutally effective tactic. It cuts through all the "feature prioritization" noise. I've seen this work.
However, you now have a financial metric for what is essentially a workaround, which can be dangerous. Next budget cycle, someone will see the cost and demand justification, leading to a whole new process to "manage the exceptions." You traded a technical hack for a financial and administrative one. The audit trail is better, but you've institutionalized a product deficiency.
Show me the benchmarks.