1. Proposed title of this feature request
Enabling secondary scheduler in openshift
What is the problem ?
Default Kubernetes scheduler (OpenShift uses default k8 scheduler) does not solve all customer use cases
Batch jobs
Prioritization of batch
Budget allocation
Priority class
For this, partners and or the upstream community are developing specialized Kubernetes schedulers (called ‘secondary schedulers’ in the rest of the document) that are well suited to solve a specific problem that the default k8 scheduler does not address.

What is the goal ?
To increase the OpenShift adoption and to be relevant in supporting next generation applications. We want to allow internally developed, partner developed, or community developed Kubernetes scheduler plugins to integrate with OpenShift in a consistent manner. Example of some of the scheduler plugins that are potential candidates
IBM LSF ( developed and maintained privately by IBM )
Volcano ( develop by CNCF community)
Trimaran (Upstream project developed in joint effort by IBM and Red hat )
Topology Aware Scheduling ( developed by upstream lead by Red Hat)

What are the guiding principles ?
Protect the default scheduler from unexpected problems created by the secondary scheduler.
Allow community/partners/vendors to develop a specialized scheduler that can be run as a secondary scheduler in openshift. Needs to co-exist with the primary scheduler
Customer should be able to run different scheduler for different workload type with in a cluster
The secondary scheduler should be
Easy to distribute
Easily to install with minimum number of steps by openshift administrator
Easy for user to change configuration in Day 2
The packaging and distribution of these scheduler will be so easy or universally known that Red Hat will not get involved
If needed we can provide an initial guide for developer on how to package and distribute
If needed we can provide an initial guide for customer on how to install
Partner/vendor/community will provide support for their own scheduler plugin
If call goes to L1 support , L1 knows how to escalate to partner
We are able to collect customer adoption statistics of these schedulers