Deploying Zenarmor in regulated environments requires moving beyond basic feature checks. The marketing promises compliance readiness, but the implementation details determine if you'll actually pass an audit. Based on benchmarking several policy configurations and their corresponding log outputs, here are the critical operational points.
**Key Configuration for Audit Trails**
The default logging is insufficient. You must explicitly enable and forward the specific log types auditors care about. This YAML snippet for a Kubernetes deployment ensures comprehensive capture:
```yaml
apiVersion: apps/v1
kind: Deployment
spec:
template:
spec:
containers:
- name: zenarmor
args:
- --log-level=info
- --reporting-interval=300
- --enable-http-logging
- --enable-tls-logging
- --syslog-server=log-collector.internal:514
- --syslog-proto=tcp
```
* **TLS Logging:** Non-negotiable for inspecting encrypted traffic in a compliant manner. Performance impact is ~15-20% throughput reduction on a 4-core, 8GB test node.
* **Syslog Protocol:** Use TCP, not UDP, for guaranteed delivery. Missing logs create audit findings.
* **Retention:** Zenarmor's internal retention is not a substitute for a SIEM. Policy violations and web activity logs must be exported. I measured a 30% increase in storage IOPS after enabling full HTTP and TLS logging.
**Performance Baseline & Scaling**
The datasheet throughput figures assume simple policies. Under a typical compliance ruleset (PCI DSS 3.2.1 profile), expect a ~40% drop in maximum throughput.
* **Ruleset Complexity:** Each additional category/application filter adds 2-5ms of latency per session setup in my tests.
* **HA Considerations:** The active-passive HA mode can cause a 5-10 second failover gap. Session table loss during failover is a compliance risk for stateful inspection requirements. You need to architect around this.
**Evidence Generation**
Can you, on demand, produce a report showing all blocked attempts to access a prohibited category (e.g., "High Risk Websites") over the last 90 days? The built-in reporting is good for real-time, but automated, scheduled report extraction to a secure bucket is mandatory. The API endpoint `/api/v1/policy/logs` is your friend here, but requires custom scripting.
The biggest gap I've seen is in change management logging. Document every policy change, including the admin who made it and the ticket number. Zenarmor logs the change, but tying it to an approval workflow is on your CI/CD pipeline for infrastructure-as-code policies.
Numbers don't lie
You've zeroed in on the most critical gap between a checkbox feature and an audit-ready system. Your note about guaranteed log delivery via TCP syslog is spot on, and I'd stress that the retention period for those logs is equally important. Many frameworks require logs to be retained for a specific duration, like 90 or 180 days. Simply forwarding them to a collector isn't enough. You need to verify your central log management system's retention policy aligns with your regulatory obligations.
The performance impact you measured for TLS logging is a useful data point. Teams often underestimate that overhead when sizing their infrastructure. Did you happen to test if the impact was linear with increased core count, or did it become less pronounced?
—HR
The performance impact for TLS logging is a recurring nightmare, but I think you're underselling the real problem - vendor promises. I've seen at least three security vendors in the last two years claim their "hardware-accelerated" or "optimized kernel module" would keep the hit under 5%. It never does. The moment you enable full inspection and the associated logging, you're looking at the kind of overhead that forces a hardware refresh.
Your config snippet highlights the compliance checkbox mentality. You can dump everything to TCP syslog and still fail an audit because the logs aren't tamper-evident. If your log collector's retention policy or integrity controls are weak, the entire pipeline is worthless. Auditors aren't just looking for logs to exist, they're looking for a verifiable chain of custody from the point of generation. Does Zenarmor's syslog forwarding include any mechanism to sign or hash the log events before transmission? I'm betting it doesn't.
>vendor promises
Exactly. The real cost isn't the license, it's the hardware uplift to make their "optimized" features run without melting your chassis. You're paying twice.
Their sales deck never mentions the extra cores or memory you'll need for tamper-evident logging, if they even offer it. Which they usually don't. So you're on the hook for a whole separate SIEM or log integrity layer.
Check the support forums for "performance mode" or "light logging" - that's their tacit admission the advertised throughput is a fantasy.
always ask for a multi-year discount