python - Google App Engine - Use Task Queues or Deferred Jobs -

google app engine has 2 methods running jobs @ later point, task queues , deferred jops

they support same features far can tell (e.g. deferred job can placed on particular task queue can throttle execution) - deferred jobs easier implement , more flexible.

anyone know of pro's con's of each method? circumstances want use task queues on deferred jobs?

i'm not sure if noticed it, documentation deferreds has section in end:

you may wondering when use ext.deferred, , when stick built-in task queue api. here our suggestions.

you may want use deferred library if:

• you use task queue lightly.
• you want refactor existing code run on task queue minimum of changes.
• you're writing 1 off maintenance task, such schema migration.
• your app has many different types of background task, , writing separate handler each burdensome.
• your task requires complex arguments aren't serialized without using pickle.
• you writing library other apps needs background work.

you may want use task queue api if:

• you need complete control on how tasks queued , executed.
• you need better queue management or monitoring deferred provides.
• you have high throughput, , overhead important.
• you building larger abstractions , need direct control on tasks.
• you webhook model better rpc model.

naturally, can use both task queue api , deferred library side-by-side, if app has requirements fit both groups.