A charger speaks to one backend. That single fact creates most of the lock-in in charging: changing platform means reconfiguring every unit, and serving two platforms from one fleet is not possible at all.
A routing layer sits between them and removes both constraints, at the cost of a component you now operate.
What routing makes possible
- Changing backend without touching charger configuration.
- Running different parts of a fleet against different platforms.
- Serving a customer's chosen platform while retaining your own visibility.
- Failing over when a backend becomes unreachable.
- Migrating in waves, with immediate rollback per wave.
The migration argument is the strongest
Without routing, a platform migration is a configuration change on every unit and rollback is the same work again. With it, migration is a routing rule and rollback is immediate.
That alone frequently justifies the layer for any operator who expects to change platform once, which is most of them.
For charger OEMs it is a product feature
An OEM whose customers each use a different platform can ship one product that connects wherever the customer needs, rather than maintaining configurations or firmware variants per customer.
It also lets the OEM retain diagnostic visibility into their own hardware while the customer's platform handles commercial operation, which is difficult to arrange otherwise.
Failover is more limited than it sounds
Routing can move connections to a standby backend, but transaction state does not follow. Sessions in progress at the moment of failover may not be recoverable on the other side, and authorisation may differ.
Failover therefore protects availability for new sessions rather than continuity for existing ones, and describing it as seamless overstates what is achievable.
What the layer costs you
A component in the path of every charger connection, which must be at least as available as the backends behind it. If it fails, the whole fleet is offline regardless of platform health.
It also adds a place where protocol behaviour can be altered, deliberately or otherwise, which means it becomes part of every integration investigation.
When it is not worth it
A small single-platform fleet with no expectation of change gains little and adds an operational component. The value appears with scale, with multiple platforms, or with a credible prospect of migration.
Adding it later is possible and means reconfiguring the fleet once, which is exactly the work it exists to avoid.