Approximation of Generalized Processor Sharing With Interleaved Stratified Timer Wheels
- Martin Karsten
- IEEE/ACM Transactions on Networking
The main objective of this project is to create a new packet scheduling technique. Packet scheduling is one of the decision process. It is used to choose which packets should be serviced or dropped. The packet scheduling is needed in network for providing maximum bandwidth, a minimum delay, a minimum packet loss rate at all times. The issues in packet scheduling techniques are space, time complexity, and tight guarantees about packets. To solve the issues in packet scheduling this project uses one method. This method is called Efficient Fair Queuing (EFQ). In this method the space and time complexity is reduced by using the flow grouping and timestamp rounding respectively. This method use a new O(1) scheduler that provides near-optimal tight guarantees. The algorithm has no loops. It has simple data structures and instructions involved makes it well suited to hardware implementations. The execution time is within two times that of DRR and consistently about three times faster than S-KPS. Speed does not sacrifice service guarantees: the WFI of EFQ is slightly better than S-KPS. From the experimental result shows that the EFQ performs well than the other existing methods.