Hey folks. I wanted to share a recent experience that really threw a wrench in our production pipeline and exposed some critical gaps in support responsiveness.
We had a Sev 2 outage last Thursday around 2 AM where our Versa SASE gateways in a critical region started dropping over 50% of packets. The control plane showed "healthy," but the data plane was clearly broken. Our monitoring screamed, and we opened a ticket immediately, marking it Sev 2 as per their own SLA definitions (significant business impact, service degraded).
What followed was… underwhelming.
**The Timeline:**
* **T+0:** Ticket opened with full diagnostics (gateway logs, packet captures from our apps, traceroutes).
* **T+30 min:** Auto-acknowledgment received.
* **T+2 hours:** First human response: "Please provide the logs and captures we already sent."
* **T+3 hours:** We re-sent everything, asked for escalation.
* **T+5 hours:** Support asked us to run basic connectivity tests from the gateways (which we'd already documented as failing).
* **T+8 hours:** We demanded a bridge call. Got a Level 2 engineer who had to "consult internally."
* **T+12 hours:** Issue finally identified as a faulty config push from their controller that corrupted a forwarding table. A rollback resolved it.
A **12-hour** turnaround for a Sev 2, where the solution was a config rollback they could have executed hours earlier.
The real kicker? Their post-mortem blamed a "rare edge case" and our "automated provisioning system" for not having adequate validation—which is ironic, because our Terraform pipeline just consumes their API. The config was valid according to their own schema.
```hcl
# Example of the "problematic" declarative config we pushed via their API
resource "versa_sase_routing" "primary" {
gateway_group = var.gg_id
static_routes = [
{
prefix = "10.10.0.0/24"
next_hops = [
{
ip_address = "192.168.1.1"
interface = "eth0"
}
]
}
]
}
```
This wasn't some complex custom rule. It's a basic static route.
I'm left wondering if anyone else has hit similar support lag on critical issues, especially when the fault seems to lie in their control plane. How do you handle escalation paths or build redundancy when the vendor's own tools and support become the single point of failure? Right now, I'm looking at architecting a multi-vendor failover, which is a cost and complexity I hoped to avoid with a unified SASE platform.