Starting point: a group plant and the questions from IT
The plant described here manufactures machines and belongs to a global group. Information security policy is set centrally, and the local IT team is responsible for the server room, the LAN and the connection to the group. A workplace safety system that will cover visitors, work permits and the location of people therefore goes through the same review as any new system on this network.
This is a deployment scenario prepared at the offer and demo stage. We do not claim that the installation is already running. We describe the assessment path, because it decides whether the project passes the IT gate before anyone starts discussing features.
- Who supplies the machines, disks and network?
- Where do the data and the backups sit?
- Which way does traffic from phones and from the vendor's service access come in?
- Can an update be tested before it reaches production?
Division of responsibility: infrastructure on the plant side
In a customer installation the infrastructure, meaning machines, network and disks, is supplied by the customer. The software vendor deploys the application in Docker containers in two instances: production and a preproduction twin. This way the IT department buys nothing it does not know, and it can apply its own standards for the operating system, virtualisation and monitoring.
InnerWeb is a software vendor, not a device maker. The installation does not require its own access control readers or the replacement of the plant's existing system. The location works without GPS and without an external map service, so there is no need to open traffic to external data providers.
Example assessment time on the plant side
Four network zones in an on-premise installation
The zone pattern structures the conversation with the network administrator. Zone 0 is the plant area with BLE beacons. Zone 1 is the users in the plant, meaning the phones of employees and visitors. Zone 2 is the plant LAN where the server stands. Zone 3 is the customer's private cloud, meaning its own virtualisation layer or data centre.
This split lets each flow have its own firewall rule: the BLE beacons do not talk to the server by IP, phones connect to the application through the entry point, and the server does not reach the internet beyond the agreed addresses.
Zones 0 and 1: site and users
The beacons broadcast a signal, and the phone calculates its position from the signals of several beacons. The server receives ready events from the application, not raw radio traffic.
Zones 2 and 3: server and the customer's cloud
The production and preproduction containers run in the plant LAN or in the private cloud. The plant makes that choice, because the same container package runs in both variants.
One entry point instead of many open ports
Traffic from outside, meaning phones on the GSM network and the vendor's service access, reaches the system only through an entry point under the customer's control. It consists of a WAF application firewall, one published address, an outbound firewall and the end of an IPsec tunnel on the customer's side. The network department therefore manages a single path that it knows and can audit.
The vendor's service access has no separate back door. It uses the same tunnel, and the plant can switch it off with one action. The details of the configuration are described on a separate page about the WAF and the IPsec tunnel in our series of implementation notes.
Data and backups stay on the plant's disks
For the group's IT team this is usually the deciding question. In this variant the data and the backups stay on the customer's disks, and the data does not leave the plant. The plant applies its own retention policy, disk encryption and restore tests.
A preproduction twin makes the most sense here: a new version first goes to an instance with the same configuration, and only after it has been checked does the IT team approve a window for production.
How the assessment could look step by step
The sequence below is an example calculation. We assume one administrator on the plant side, ready virtual machines and no unusual requirements from the group. The times are indicative and depend on the change queue at the plant.
- Review of the architecture and zones with the network administrator: 1 working day.
- Agreeing the firewall rules and the tunnel: 2 working days.
- Installing preproduction and testing the login: 1 working day.
- Restoring a backup on preproduction: 1 working day.
- Decision on the production instance: 1 working day.
Conclusions for the plant's IT team
First, the assessment starts with the division of responsibility, not with features. Second, four zones and one entry point shorten the conversation with the group's security department, because every flow has its own rule. Third, preproduction moves the risk of updates from production to a test environment. It is worth comparing this scenario with your own checklist for accepting an installation.
Video
Frequently asked questions
Who buys the server for the system?
The customer supplies the infrastructure, meaning machines, network and disks. The software vendor deploys the application in Docker containers in the production and preproduction instances.
Does the data leave the plant?
In the described variant the data and the backups stay on the customer's disks. The data does not leave the plant, and traffic from outside passes through one entry point.
Does the system require our access control readers?
No. InnerWeb is a software vendor and does not require its own readers or the replacement of the plant's existing system.
Does the location need GPS or external maps?
No. The location works without GPS and without an external map service, which simplifies the outbound firewall rules.
Sources
Let's walk through the architecture with your IT team
We will present the split into zones, the entry point and the preproduction test during a technical meeting with the plant's IT team.
Let's talk