Scheduler Architecture Within Existing Infrastructure

Nowadays, schedulers are used in every company trying to optimize their datacenter utilisation and leverage other resources such as AWS, Azure, Softlayer, GCP. Multiple tools are available from proprietary software such as Control-M (BMC), Autosys (CA), Dollar Universe (Automic), One Automation (Automic) to open source solutions such as ProActive (ActiveEon).

Regarding installation of this software, customers have the choice between: - Having one scheduler managing all resources, - Having two schedulers and split the resources, - Or having more schedulers. In my opinion, the decision can depends on parameters such as number of environments, IT policies, component isolation, high availability requirements, etc.

What is the architecture used in your company? What are the reasons behind your choice? What was the challenges behind this decision?