By Jeffrey Mcdaniel-Oracle on May 03, 2013
As you know the Extended Schema services (Project service specifically) runs on a designated time interval based on your Publish Project settings. This could be set to run every 1 min (default), and at this time the 'arbiter' runs and determines which projects need to be queued up to be processed. The arbiter determines this by evaluating each project that has PX_ENABLE_PUBLICATION_FLAG set to 'Y', and threshold of changes on this project or the threshold of time has been crossed. In your database you can see how many changes there are on the specific project by reviewing your PLPROJREF table in the Admuser schema.
Another reason why a project would be queued up is because you right clicked on the project and choose Publish Now. When you choose publish now what actually occurs is a time stamp is made on the PX_NEXT_DATE field for that project row in the PROJECT table. In the application there is not a way to separate out projects into different batches that may follow different publish project settings. The design behind this was to allow data to continually be updating in the system at these different times when the project may have an appropriate amount of changes, time has passed, or you need it now then to calculate the project. Let the system take care of itself and keep the data up to date in a near real time environment. On the P6 Analytics\P6 Reporting Database side controls were added such as filters, and additional data sources to allow you to control what projects get pulled over and possibly setting them up on different ETL runs.