Who can access the OutPost VM?

Only the OverView platform, through its private network. The VM is reachable neither from the Internet nor from another OverView customer.

Can OutPost reach my whole internal network?

No. By default it has access to nothing. Every internal API must be explicitly opened by a rule on your firewall, to a specific IP and port.

What happens if the VM is compromised?

An attacker would only see the APIs you allowed, read-only. OverSOC revokes the VM’s access in under 30 seconds, and you can delete it with zero impact on your systems.

Do I need a public IP or a bastion?

No. The VM dials out through your NAT and firewall like any workstation, directly when the network allows it and through an encrypted relay otherwise. Nothing to expose, nothing to publish.

What is the performance impact?

Negligible. WireGuard overhead is under 5% and connections go direct whenever the network allows.

Do I need an agent on my servers?

No. OutPost queries the existing APIs of your tools. Nothing to install on your production servers.

How much of my network team’s time does it take?

About an hour: 15 minutes to import the VM, 5 for the network, 15 to 30 for the firewall rules. The “hand this to your network team” block above has everything they need.

What if my tool has no API?

Use OverView DropPoint: a dedicated S3 drop-off space where you push CSV or JSON exports, manually or from a scheduled script.

ON-PREM GATEWAY · OVERVIEW OUTPOST

Your internal tools feed your metrics. OutPost installs in an hour, with no inbound port to open.

Q: Is it a VPN?

Technically yes: OutPost ships with a WireGuard tunnel (Tailscale mesh network), pre-configured and operated by OverSOC. What you avoid is the site-to-site VPN project: no concentrator to configure, no routes to advertise, no keys to manage. The VM dials out, authenticates, done.

GLPI, vCenter, Active Directory, SCCM, Nessus: the data that matters sits behind your firewall. OutPost is a virtual machine you deploy in your DMZ. The encrypted WireGuard tunnel is built in: the VM opens it outbound on its own, and nothing comes back in, and it relays only the APIs you allow, one firewall rule per API.

Start with your on-prem tools Deployment guide ↗

Encrypted WireGuard tunnel, opened by the VM
No inbound connections from the internet
One firewall rule per API, nothing else
Why it is safe ↓

WHAT CHANGES

0VPN to build, 0 NAT, 0 agent

≈ 1 hfrom importing the VM to the first API read

1 ruleper allowed API on your firewall, nothing else

< 30 sto revoke the VM’s access

THEY DID IT

Clésence connected AD, GLPI, EDR and monitoring without rebuilding its inventory

A social housing company in the Action Logement group, multi-site infrastructure, outsourced CISO. OverView read the existing inventories through their APIs, and their CMDB is now fed back automatically with consolidated data.

Read the case study →

Alexandre Demol

Infrastructure Manager — Clésence

“For every incident, we spent 2 to 3 hours rebuilding what happened. With OverView, the information is right there — we went from reactive to preventive.”

THE PROBLEM

Your most useful data sits behind your firewall

One VPN per vendor

Every SaaS tool that wants to read your GLPI, your vCenter or your Active Directory asks for a site-to-site VPN, a security review and weeks of network tickets. Your reporting project waits on the network project.

Open ports you end up regretting

Exposing an internal API to the Internet, even behind an IP filter, is one more attack surface to justify at every audit.

Agents everywhere

Installing an agent on every server to collect data means rollout, updates and exceptions to manage forever.

HOW IT WORKS

An outpost, not a front door

An outpost is a small forward position, set at the edge of the territory, that observes and reports back to base. It commands nothing, stores nothing, and can be withdrawn without affecting the main position.

01
Deploy the VM
Import the provided image (OVA, VHDX or QCOW2) into your hypervisor, in a DMZ or isolated VLAN. A lightweight VM, about 15 minutes.
02
Allow outbound traffic
The VM initiates a WireGuard tunnel to OverView itself. On your firewall: outbound HTTPS and UDP, no inbound rule, no port forwarding.
03
Open only what is needed
By default OutPost sees nothing on your network. You add one rule per API to query: the GLPI IP on 443, the vCenter IP, nothing else.
04
Plug in your connectors
OutPost appears in OverView as a connection route. You configure your internal sources exactly like a cloud connector, and your metrics are recomputed on the next sync.

WHICH ROUTE FOR WHICH TOOL

Three routes, one inventory

Cloud, on-prem, file: every source has a route into OverView. Most customers combine two of them.

Your situation	Recommended route
SaaS tool with a public API (EDR, Entra ID, Intune, ServiceNow)	Cloud API
Internal tool with an API (GLPI, vCenter, Active Directory, Nessus, Centreon)	OverView OutPost
Several internal tools behind the same firewall	OutPost, one VM for all of them
Tool with no API, or that only produces exports (scanner report, antivirus console, home-grown CMDB)	OverView DropPoint
Isolated site or a network you do not open	DropPoint, a file goes out, nothing comes in
Both cases in the same infrastructure	OutPost + DropPoint, same inventory

SECURITY

The answers your security review will need

Outbound connections only

The VM initiates every connection and crosses your NAT like a workstation would. No public IP, no bastion, no reverse proxy, no internal API exposed.

End-to-end encrypted

WireGuard (ChaCha20-Poly1305) with automatic key rotation. Any relay only ever sees encrypted traffic.

Zero trust by default

At install time OutPost has access to nothing. Every internal API must be explicitly allowed by your firewall, IP and port included.

Dedicated private network

OutPost joins OverView’s private network only. The platform alone can reach it, never another customer or the Internet.

Limited impact if compromised

If the VM were compromised, an attacker would only see the APIs you allowed, read-only. Its access is revoked in under 30 seconds.

Hardened base, self-updating

Hardened OS with automatic security updates. Nothing for you to maintain.

Recommended reading before your security review

HAND THIS TO YOUR NETWORK TEAM

What the firewall must let out, and nothing else

Paste this block into the ticket. The deployment guide details the rules for FortiGate, Palo Alto, Check Point, Cisco ASA / Firepower and pfSense / OPNsense.

01Outbound TCP 443 and UDP 3478 from the VM to the control plane and relays (*.tailscale.com)
02Outbound high-port UDP from the VM to the OverView egress IP, for the direct WireGuard connection
03Outbound TCP 80 / 443 to the Ubuntu and Tailscale repositories (updates), UDP 123 to ntp.ubuntu.com
04Internal: VM IP to each target API on its port (e.g. 443), everything else denied by default
05Stateful firewall, no SSL inspection on this traffic, no inbound rule

Detailed rules per vendor for your firewall ↗Talk to an OverSOC engineer →

TECH SHEET

What to plan for

Sizing: 2 vCPU · 2 GB RAM · 10 GB disk
Hypervisors: VMware vSphere / ESXi (OVA), Microsoft Hyper-V (VHDX), Nutanix AHV, Proxmox VE, KVM (QCOW2)
OS: Hardened Ubuntu 24.04 LTS, automatic security updates
Network: DHCP by default or static IP, internal DNS resolution supported
Outbound traffic: TCP 443 and UDP 3478 to the control plane, high-port UDP to OverView
Inbound traffic: None
Placement: DMZ or isolated VLAN, behind a stateful firewall
Documented firewalls: FortiGate, Palo Alto, Check Point, Cisco ASA / Firepower, pfSense / OPNsense

WHAT YOU UNLOCK

Your internal tools become OverView sources

CMDB, hypervisors, scanners, directories, monitoring, backup: as soon as a tool exposes an API on your network, OutPost lets OverView read it. One VM covers every tool behind the same firewall, added one at a time, with nothing installed on them. And every source unlocks metrics.

GLPI

iTop

EasyVista

VMware vCenter

Nutanix

Tenable Nessus

Tenable Security Center

OpenVAS

Cyberwatch

Active Directory

Wallix

Centreon

Zabbix

PRTG Network Monitor

LibreNMS

Veeam

SCCM / MECM

WSUS

Lansweeper

NetBox

EfficientIP SOLIDserver

Examples of metrics unlocked

EDR coverage of servers, against the real inventorySources: vCenter + EDR
Devices active in AD but unknown to your ITSMSources: Active Directory + GLPI
Obsolete servers with High vulnerabilitiesSources: Nessus + vCenter
Percentage of VMs backed upSources: vCenter + Veeam
Active / inactive AD accountsSources: Active Directory

See all integrations →

FREQUENTLY ASKED

What your CISO will ask