Perclient

Perclient (per-client) is a design principle in software systems that emphasizes applying configuration, policy, and behavior on a per-client basis within a multi-tenant environment. It supports isolation between clients while allowing individualized customization.

Common applications include rate limiting per client, feature flags, data access controls, quotas, billing, logging, and auditing. By using per-client scopes, operators can tailor service levels and capabilities for each client without affecting others.

Implementation typically relies on identifying clients through keys or identifiers (API keys, client_id, tenant_id), storing per-client settings in a database or configuration service, and injecting those settings into request handling paths. Systems may implement per-client routing, authorization, caching, and data partitioning. Performance considerations include caching per-client policy and minimizing lookups.

Challenges include configuration complexity, consistency across services, versioning, and secure isolation to prevent cross-client data leakage. Testing per-client behavior requires careful planning, and rollout often uses feature flags and staged deployments.

See also per-tenant, per-user, multi-tenant architecture, and policy as code.