Processor Sharing
Processor Sharing (PS), Egalitarian Processor Sharing (EPS), and Discriminatory Processor Sharing (DPS) (DPS) are s primarily used in queuing systems, especially in the context of computer networks and systems. Here's a brief overview of each:
Processor Sharing (PS):
In a PS system, all jobs (or packets or tasks) in the system are served simultaneously.
Each job receives an equal fraction of the service capacity. For instance, if there are n jobs in the system, each job receives 1/n of the server's capacity.
PS is idealized and may not be practically implemented in many real-world systems, but it serves as a useful theoretical model for analyzing system performance.
Egalitarian Processor Sharing (EPS):
EPS is a variant of PS where all jobs of the same class are treated equally.
In systems with multiple classes of jobs, each class gets an equal share of the processor, and within each class, jobs are served using PS.
For example, if there are two classes of jobs and each class has multiple jobs, each class gets 50% of the server's capacity, and within each class, the jobs share the class's capacity equally.
Discriminatory Processor Sharing (DPS):
DPS is another variant of PS where different classes of jobs receive different fractions of the server's capacity based on predefined weights.
Each class is assigned a weight, and the server's capacity is divided among the classes based on these weights.
Within each class, jobs share the class's capacity equally (like PS).
DPS allows for differentiated service levels for different classes of jobs.
In summary, while PS serves all jobs equally, EPS and DPS introduce the concept of classes and allow for differentiated service levels based on class membership. EPS ensures equal service for each class, while DPS allows for weighted service based on the importance or priority of each class.