Blue Prism Technical Architect flashcards

32 cards. Read the term and the definition here. The deck at the top of this page is optional.

Application Server role

A Windows service that brokers connections between clients, runtime resources, and the database. It does not execute process logic itself; resources do that.

Runtime resource

A licensed machine with an interactive session that executes published processes. It cannot edit process definitions, which separates build from run.

Blue Prism database

The central store for process definitions, work queues, logs, credentials, and audit records. It is the single source of truth for the environment.

Interactive client licence

A licence consumed by a user who builds, edits, or controls processes. It is distinct from a runtime resource licence used for execution.

Centre of Excellence

The governing body that owns automation standards, reusable assets, change control, and pipeline decisions across the organisation under the operating model.

Controller role

The run-time owner who monitors queue health, triages exceptions, and escalates incidents. The controller does not normally change process logic during live running.

Process pipeline scoring

A weighted assessment of candidate processes using volume, standardisation, exception rate, data quality, and regulatory exposure to prioritise automation investment.

Process Design Instruction

A pre-build document capturing process steps, systems touched, inputs and outputs, exception paths, and volume profile. It is the design baseline for development.

Producer-consumer pattern

A design where one process creates queue items and separate consumer processes handle them. It decouples intake rate from processing capacity.

Transaction boundary

The smallest unit of work that can be completed and recorded independently. It usually maps to a single work queue item.

Environment variable usage

Environment-specific values such as URLs, file paths, and credential references are stored here rather than in process logic, allowing unchanged promotion between environments.

Block stage recover pattern

A group of stages wrapped with a recover and resume pair so the whole group can be retried as a unit before the exception is rethrown.

Stage description standard

Every stage carries a description of its purpose. This is a review requirement because it makes logic auditable without reading every expression.

Object action idempotency

Repeating an action must not create duplicate records or side effects. Check-before-write logic inside the object enforces this.

Attach mode

The rule that binds an object to a running application instance, for example by process name, window title, or handle. Correct mode prevents orphaned or wrong-target automation.

Stable element attributes

Element matching uses automation identifiers, names, and classes rather than screen coordinates or ordinal position, which break when layout changes.

Spy mode selection

The technique used to capture elements: Win32 for classic controls, UIA for modern frameworks, browser modes for HTML, with region or image as fallbacks.

Work queue item status

Items move through New, In Progress, Exceptioned, and Completed. Deferred items return to the queue after a configured interval for retry.

Queue encryption

Sensitive fields stored in queue data are encrypted so that database-level access does not expose personal or confidential information.

Collection persistence limit

Collections live in memory and are lost when the session ends. Anything that must survive a restart belongs in a queue or external store.

Credential store

The protected repository holding named credentials with username and password. Processes reference credentials by name and never embed secrets in logic.

Role-based access control

Permissions are granted to roles and users per folder or process. Least privilege applies to designers, controllers, and runtime service accounts alike.

Audit log purpose

Records logins, definition changes, and credential access. It provides the evidence trail regulators and internal audit require.

Business versus system exception

A business exception means the data is invalid and the item should stop. A system exception is a transient fault that warrants bounded retry.

Bounded retry with backoff

A fixed maximum attempt count combined with an increasing wait interval. It prevents infinite loops and reduces load on a failing target system.

Stale in-progress recovery

A routine that finds items left In Progress by a crashed resource and returns them to the queue so work is not silently lost.

Environment promotion

Moving a packaged release of processes, objects, and dependencies from development through test to production, with the same logic and different environment variables.

Release rollback plan

A documented path back to the prior known-good package, including the trigger conditions and the steps to restore it during a failed deployment.

Pre-import dependency comparison

Checking the target environment for missing objects, credentials, and environment variables before import, to avoid a partially working release.

Throughput per resource

Available run seconds divided by average handling time per item. Total capacity multiplies this by the number of licensed resources.

Bottleneck tier identification

Determining whether UI interaction, database, API rate limits, or network is the constraint before adding resources, since saturation elsewhere makes extra bots useless.

Capacity headroom margin

Extra capacity planned above average daily workload to absorb peak volume, retry overhead, and rework of exceptioned items without breaching service levels.