← Course Home
Okta Certified Workflows - Specialty · Part I

Building Flows

43% of Part I — the exam's largest section, split across 4 modules.

Building Flows · Module 3

Sub-topic: Data Streaming

this deck covers 1 of 4 sub-topics in this module

↑ Module 3 Overview

The Scenario

"Set up data streaming on an action card to search for or list records"

A flow needs to process a data set that's far too large to handle normally — well beyond the 10,000-record ceiling on ordinary search results. This is the mechanism the exam expects you to reach for.

The Mechanism

The Stream Matching Records option appears on many action cards that have search or list options — for example, Freshservice Search Tickets or Okta List Users with Filter. Instead of processing the data set inside the parent flow, you point the card at a helper flow (via the Choose Flow dialog) that's built to handle many records. Using pagination-enabled API endpoints and a helper flow, a flow can process up to 1 million records.

Why It Exists

Even where search-result limits are respected, wide records — many custom user properties or heavy user data — can still push a flow past system limits and trigger out-of-memory errors. Streaming shunts that processing load off into a dedicated helper flow so the parent flow doesn't choke on it.

Performance Benchmarks

Processing 500,000 records: ~13 hours
Processing 1 million records: ~25 hours
Workflows may process flows simultaneously (concurrency) to finish within a reasonable time

These are real published benchmarks, not estimates — expect the exam to test whether you know streaming is for genuinely large jobs, not a casual default.

The Angle — What Trips People Up

Watch for this in your own sandbox run:

→ Assuming you can stop a streaming flow mid-run. You can't — once execution begins, a helper flow that streams data runs until the API returns all records or the flow hits its specified maximum. An in-progress flow doesn't even stop if you deactivate the flow.

→ Forgetting the hard ceiling: the maximum limit of returned records is 1 million, and processing halts when it reaches the page limit you (or the connector) specify.

One Line To Remember

Data streaming replaces in-flow processing with a helper flow driven by pagination, scaling a search/list card from the 10,000-record search cap up to 1 million records (~13 hrs for 500K, ~25 hrs for 1M) — but once it starts, you can't stop it, and deactivating the flow won't halt an in-progress run.

Source: help.okta.com — Stream data with action cards (ext-about-streaming)