Informatica Cloud DI Foundation key terms

A lightweight, self-updating runtime installed on a machine or container that executes mappings, tasks, and connectors on behalf of the cloud-hosted Informatica organization. It pulls job instructions from the cloud and processes data locally or in the customer's network.

A logical collection of Secure Agents that share configuration and can run the same jobs. Assigning a task to a group lets the service pick an available agent in that group, which supports load balancing and high availability.

The combination of a Secure Agent or agent group, an operating system, and optionally a compute cluster that determines where and how a job executes. Tasks reference a runtime environment to select execution capacity.

A cloud-managed execution option where Informatica provisions and scales compute for jobs, removing the need to install and maintain Secure Agents yourself. It is typically used for elastic or short-lived workloads.

The top-level tenant in the Informatica Cloud service that contains users, roles, connections, assets, and runtime environments. Sub-organizations can be created beneath it to separate projects or business units.

A child tenant under a parent organization that has its own users, assets, and licenses while being administered from the parent. It is used to isolate environments such as development, test, and production.

A named set of privileges assigned to users that controls which assets they can view or change. Roles range from administrator, who manages the organization, to designer and operator roles limited to building or monitoring jobs.

A saved definition of how to reach a data source or target, including connector type, host, credentials, and runtime environment. Mappings and tasks reference connections instead of embedding access details.

An individual setting inside a connection, such as host name, port, user name, password, or schema. Some properties can be exposed as parameters so the same connection definition works across environments.

A reusable design that defines how data moves and transforms from sources to targets. It contains transformations, expressions, and links, and is stored as a versioned asset that tasks can execute.

The runnable object that binds a mapping to specific connections, a runtime environment, and scheduling or trigger settings. Separating design from execution lets one mapping serve many environments and schedules.

The mapping object that reads records from a source connection or object and passes them downstream. It defines the source object, the connection, and which fields are projected into the data flow.

The mapping object that writes incoming records to a target connection or object. Its properties control operation type such as insert, update, upsert, or delete, plus error handling behavior.

A reusable processing component placed in a mapping that changes, filters, joins, aggregates, or routes data. Each type has defined input and output ports and configurable properties that determine its behavior.

A transformation that computes new or modified port values row by row using expressions, functions, and variables. It is commonly used for data cleansing, concatenation, type conversion, and deriving keys.

A transformation that retrieves related values from a lookup source, such as a table or file, and returns them to the data flow. It supports caching and can be configured to return multiple matches or a default value.

A lookup transformation that is not linked into the main data flow and is instead called from an expression by name. It returns a single value per call, which suits point lookups inside calculated fields.

A transformation that combines two data streams based on matching key fields, supporting inner, left, right, and full outer joins. It also distinguishes master and detail pipelines and can use sorted input for efficiency.

A transformation that groups rows by key and computes summary values such as sum, count, average, minimum, and maximum. Its group-by ports define the grouping, and sorted input can improve performance.

A transformation that evaluates conditions and sends each incoming row to one or more output groups, plus a default group for unmatched rows. It replaces multiple filter branches with a single component.

A transformation that passes only rows meeting a specified condition and discards the rest. It is placed early in a pipeline to reduce downstream processing volume.

A transformation that produces sequential numeric values for each row, optionally restarting per session. It is used to create surrogate keys when the target does not generate them.

A transformation that tags each row with an operation code such as insert, update, delete, or reject so the target applies the correct action. It enables mixed-operation loads from a single pipeline.

A write operation that inserts a row when no matching key exists and updates it when a match is found. Behavior depends on the target connector's support and the key fields configured on the target.

The practice of replacing hard-coded values in assets with parameters resolved at run time, often from a parameter file or task input. It lets one design run against different environments and date ranges.

A file that supplies values for parameters referenced by mappings or tasks at run time. It centralizes environment-specific settings so designs do not need editing between development and production.

A configuration that starts a task automatically at defined times or intervals, such as hourly, daily, or on a cron-like expression. Schedules are tied to the task and the time zone of the organization.

The service area where users view running and completed job instances, inspect logs, see row counts, and review errors. It is the primary tool for troubleshooting failed mappings and tasks.

A detailed record of what happened during a job run, including reader and writer statistics, transformation errors, and SQL issued. Log levels can be raised temporarily to capture more diagnostic detail.

An output file that collects rows that failed to be written or processed, along with the reason for each rejection. Reviewing rejects helps correct data quality issues without rerunning the entire load.

The set of settings that determine what happens when a row or the whole job fails, such as stopping the job, skipping the row, or writing it to a reject file. It balances data completeness against pipeline reliability.

The classification of a port or field, such as string, integer, decimal, date/time, or binary, that determines valid values and conversion rules. Mismatches between source and target types cause errors or truncation.

One or more fields that uniquely identify a row in a table and are used to match records during update or upsert operations. Defining keys correctly prevents duplicate inserts and unintended overwrites.

A technique that converts transformation logic into SQL or native operations executed by the source or target database instead of the agent. It reduces data movement and can greatly improve performance on large volumes.

Splitting a pipeline into parallel streams so multiple threads or compute nodes process data concurrently. It improves throughput on large loads but requires care with ordering-sensitive transformations such as aggregators.

A load pattern that processes only new or changed records since the last run rather than the full data set. It typically relies on timestamps, change flags, or change data capture to identify affected rows.

A load pattern that reads and writes the entire source data set on each run, often after truncating the target. It is simpler to design but costs more time and resources as data volume grows.

The process of moving assets such as mappings, tasks, and connections between organizations or sub-organizations, usually through export and import. It requires mapping connections and parameters to the target environment.