MSAMM
Back to course

Oracle Integration Cloud: Building Integrations That Survive Production · Module 14 · Performance and throughput

Parallel processing, and its cost

Lesson 139 of 175 · 2 min

Running branches concurrently, the limits that apply, and THE TARGET SYSTEM THAT THROTTLES YOU FOR DOING IT. Parallelism is the obvious lever when a flow is too slow and it is the one most often pulled without measuring. Split the work, run several branches at once, finish in a fraction of the time. Sometimes that is exactly what happens. OFTEN IT IS NOT, BECAUSE THE BOTTLENECK MOVES RATHER THAN DISAPPEARING. Lesson 1's point: if the time is in the target system, running ten branches means the target is now receiving ten times the traffic. A system that answered in 300ms under one caller may answer in 3 seconds under ten, and the flow finishes in the same wall-clock time while consuming ten times the resource. AND SOME TARGETS RESPOND BY THROTTLING YOU, WHICH IS THE FAILURE IN THE TITLE. An API with a rate limit answers 429 once you cross

The full lesson is part of the course

The video, the complete written lesson and the module quiz are included in Oracle Integration Cloud: Building Integrations That Survive Production, with a certificate on completion and a fourteen-day refund window.

Get the free lessons by email

We will email you a link to every free lesson in this course. No account needed, and one message only.

In this module: Module 14 · Performance and throughput

  1. 1Where the time actually goes
  2. 2Per-record calls versus batch
  3. 3Parallel processing, and its cost
  4. 4Payload size and memory
  5. 5Message packs and cost per run
  6. 6Throttling and back pressure
  7. 7What breaks: the integration that worked in test
  8. 8Lab briefing · Ten times the volume