Just wrapped another 3 AM shift where our "zero-trust" vendor gateway decided to take a nap. Naturally, it got me thinking about Perimeter 81.
We're evaluating it to potentially replace our current hair-pulling solution. The marketing spiel sounds great—cloud-based, integrates with everything, yada yada. But I need to know what it's like when the pager goes off at 2 AM on a Tuesday.
Specifically:
* **Terraform/API maturity:** Is their IaCP story real, or just a few API endpoints bolted on? I don't want to click in a UI to onboard every new microservice.
* **K8s sidecar/egress:** Anyone running it for cluster egress control? How's the overhead?
* **The actual support:** When something breaks, do you get a script-reading drone or someone who can actually read a `tcpdump`? What's the average real fix time?
Our nightmare scenario is having to explain BGP to a L1 support agent while our production VPN is down. Been there, done that, got the burnt-coffee-stained t-shirt.
Would love some trench stories—good, bad, or ugly. Bonus points if you've had to open a Sev-1 ticket.
NightOps
The Terraform provider works for basic network and gateway objects, but it's lagging on the identity side. We've had to maintain a mix of API calls and manual UI steps for SCIM user provisioning, which defeats the IaC purity you're after.
On your Kubernetes egress question: we've been running the sidecar in a few test namespaces for about six months. The overhead is manageable, roughly 10-15ms added latency, but the real issue is the lack of native pod identity mapping. You'll end up using service accounts in a way that feels like a workaround, and their logging for denied egress attempts is too generic to be useful for debugging.
For a Sev-1, we had a regional connector fail and their support engineer actually joined a Zoom and walked through our packet captures live. That was impressive, but it took 47 minutes to get that engineer on the line. The front-line folks absolutely will ask you to reboot the gateway first.
Data over dogma