OutSystems Architecture Specialist (O11) flashcards
OutSystems Architecture Specialist (O11) · Build recall with a focused review session.
32 cards. Read the term and the definition here. The deck at the top of this page is optional.
- OutSystems canvas layers
- The four-layer dependency model: End-User (UI), Core (business domains), Foundation (cross-cutting), Library (stateless reuse), plus Service for external APIs. Dependencies point downward only.
- End-User module scope
- Contains screens, UI logic, and session-bound behavior. It may consume Core, Foundation, and Library elements but must never be referenced by them, keeping business logic reusable.
- Core module responsibility
- Owns one business domain's entities and rules and exposes public actions. It may reference Foundation and Library but must not reference another domain's Core module directly.
- Foundation module purpose
- Holds domain-agnostic cross-cutting logic such as integration adapters, utilities, and authentication helpers. It must not own business entities and sits below Core in the canvas.
- Library module constraints
- Stateless, side-effect-free reusable logic with no references to Core, Foundation, or End-User modules. Safe to share across applications because it carries no domain state.
- Entity ownership rule
- Exactly one module creates and owns each entity; all other modules access it read-only or through public actions. Prevents duplicate entity definitions and conflicting write logic.
- Circular module reference
- A producer/consumer loop between modules, blocked by the platform. It signals a layering defect: extract the shared logic into a lower-layer module both can consume.
- Module granularity heuristic
- Create one Core module per bounded business capability. Splitting by screen or CRUD verb creates chatty dependencies and inflates impact analysis and deployment coordination cost.
- Horizontal scaling in OutSystems
- Add front-end servers behind the load balancer. Session state lives in the platform database, so servers are stateless and sticky sessions are not required for normal operation.
- Front-end vs back-end zones
- Front-end servers host End-User modules for user traffic; dedicated back-end servers host Core, Foundation, and integration modules, isolating CPU-bound work from interactive requests.
- Read replica rule
- Point reporting and read-heavy queries at a dedicated read replica or reporting database, keeping interactive screens off the same connection pool as batch and analytics workloads.
- Asynchronous offload threshold
- Any screen action expected to run beyond a few seconds should be evaluated for async processing via BPT, timers, or queues, freeing the request thread for user traffic.
- Cache invalidation rule
- Cache only data read far more often than written that tolerates staleness. Invalidate explicitly on write and never store per-user sensitive data under a shared cache key.
- Integration resilience patterns
- External calls require timeouts, retries with backoff, and circuit breakers. Chained synchronous calls multiply latency, so aggregate or parallelize when the external API permits.
- Screen aggregation rule
- A screen should call one aggregate or a small fixed number of server actions. Loops that query the database per iteration create N+1 patterns, the most common scalability defect.
- Capacity headroom rule
- Plan capacity from measured throughput per server plus 30-50% headroom. Scale out before sustained peak CPU exceeds roughly 70%, so spikes are absorbed without degradation.
- Deployment pipeline stages
- Develop, Test/QA, Pre-Production, and Production, each a separate deployment zone. Code is promoted as a solution through stages, never re-created manually per environment.
- LifeTime role
- The platform console for managing applications, deployments, and environment configuration. Environment-specific values belong in platform configuration, not hard-coded in module logic.
- Impact analysis before publish
- Before publishing, review TrueChange warnings and dependency impact. Breaking a public action signature forces every consumer module to be republished and retested.
- Technical debt baseline
- Module complexity and integration indicators plus TrueChange warnings form the health baseline. Schedule refactoring as warnings accumulate, before incidents force unplanned work.
- Semantic versioning rule
- Breaking changes to public elements require a major version plus consumer coordination; additive, backward-compatible changes take a minor version. Applied to modules and solutions.
- Pipeline quality gate
- A stage passes only when automated regression and smoke tests are green and TrueChange is clean. Gates block promotion, catching defects before they reach production.
- Cross-environment performance comparison
- Compare platform performance monitoring and logs across stages. Divergence between environments points to configuration differences; consistent slowness points to code regressions.
- Reproducible rollback
- Keep the previous tagged solution version deployable in every stage, so rollback is a redeploy of a known version rather than an emergency code fix under pressure.
- Architect canvas ownership
- The architect owns layer placement, module boundaries, entity ownership, and naming, and publishes the canvas as the reference before development begins.
- Developer module autonomy
- A developer owns internal logic, local variables, and private actions within an assigned module. Changing a public element signature requires architect review.
- Public element as contract
- Anything consumed by another team - public actions, entities, exposed APIs - is a contract. Changes need the owning team's agreement, unlike private module internals.
- Team topology alignment
- Team structure mirrors module ownership: one team per set of Core modules, with Foundation and Library owned by a shared platform or services team.
- Architecture review gates
- Canvas conformance checked at design, public-element changes at code review, and dependency violations blocked in the pipeline instead of being discovered in production.
- Non-functional standards ownership
- The architect defines caching policy, async thresholds, integration patterns, naming, and error handling; developers apply them and document any approved deviation as an exception.
- Cross-domain data escalation
- When a team needs another domain's data, it uses an agreed Core public action or a Foundation service. Direct references between teams' Core modules are forbidden.
- Architect governance artifacts
- The architect maintains the canvas diagram, module inventory with owners, public-element catalog, and the technical-debt backlog prioritized jointly with product management.