This is really critical for an actual path to a big data platform which I know you are trying to become. The limit on max collections allowed to setup archive should match the max number collections allowed on a cluster. Without that, you really can not be setup to handle big data in one solution. The number of actively running jobs can be a smaller number but should be configured to rotate thru the ones setup, so that you eventually get everything needed into archive.
This is really critical for an actual path to a big data platform which I know you are trying to become. The limit on max collections allowed to setup archive should match the max number collections allowed on a cluster. Without that, you really can not be setup to handle big data in one solution. The number of actively running jobs can be a smaller number but should be configured to rotate thru the ones setup, so that you eventually get everything needed into archive.