Placement service has been introduced since Newton release, and has been rapidly evoluted for the past few cycles. Placement has been proved to be a efficient solution for scheduling instances in large scale deplyemnt. With the help of placement services, users and opertors will be also able to perform more advanced instance scheduling process that are yet unable to achieve. In this topic we will introduce the basic data model that are used in placement and how it has been used in the latest openstack release that helped users to achieve advanced scheduling such as bandwidth aware scheduling(intergration with Neutron), volume type based scheduling(intergration with Cinder), etc.
We will also cover how new projects like Cyborg can integrate with Core projects like Nova to achieve their goal in a decoupled way.
The attendees will be able to learn, how placement can help you achieve advanced scheduling process, such as:
1. How resources are modeled in placement and why is it efficient;
2. How Nova-Neutron achieve bandwidth aware scheduling with the help of placement;
3. How Nova-Cinder achieved volume-type aware scheduling with the help of placement;
4. How new projects like Cyborg can integrate with old projects like Nova with the help of placement.
