GLOSSARY
Cluster
A cluster groups multiple connected computers so they can process tasks together or provide services; the specific distribution of work depends on the system.
Shared capacity needs a clear purpose
When many users analyse documents simultaneously, a single computer may become a bottleneck. A cluster makes it possible to distribute work across several machines. That helps only if the application can divide the work sensibly and its other components can keep up. Adding application servers does not automatically accelerate an overloaded database. Customers therefore need evidence of the reliable capacity of the complete service. The number of machines is one technical factor, not an independent measure of service quality.
Resilience needs demonstration
Several machines create options for handling failures. Data, state and responsibilities still need appropriate organisation. A common configuration error or shared dependency can affect multiple components at once. Kubernetes can coordinate some operational tasks, but cannot replace sound architecture. A useful test combines realistic load with the controlled failure of a component and checks whether work continues correctly. Total operating cost includes monitoring, updates and operational support. A simpler setup may be more economical for a small pilot when it meets the actual requirements.