Architecture
Deploying MONARC with a back office requires the use of at least 4 different servers:
- a server acting as a reverse proxy (RPX);
- a configuration server (CFG) where ansible is running;
- a back office server (BO);
- at least one MONARC server (FO).
The MONARC architecture is presented here.
Only the RPX should have a public IP address and be available via HTTPS (port 443). It is advised to install a tool like Fail2ban on this server.
The administration of the servers should be done via SSH and only possible inside a VPN in order to prevent any attack (port scanning, etc.) from the outside.
Multitenancy
MONARC is a multitenant application. Every tenant (MONARC instances on a FO server) are separated:
- databases (MONARC models, users management, etc.);
- data (assets like deliveries templates, configuration files, cache, etc.);
- Apache instances with Virtual Hosts.
An administrator of a back office is not able to access to a MONARC instance. The administrator can only create or delete clients (instances). Read this section for more information.
Generally MONARC instances are virtually isolated on one FO server. It is also possible to isolate instances on different servers in the same subnetwork. Or on physically distant servers if you are using a VPN.
The reverse proxy is responsible of dispatching the MONARC users to the correct instances.
Configuration server
The configuration server (CFG) must be able to contact the other servers via SSH with public encryption key. The corresponding private key must be stored only on this server, for example for a system user ansible. This user will launch periodically the MONARC playbook with cron.
System update
Always keep your systems up-to-date. At least the security updates from the GNU/Linux distribution.