|
|
|
|
|
|
|
MOTIVATION: Huge ammounts of information are stored on tapes. Requests for data on one of this tapes arrive randomly. In order to read the data the tape has to be mounted, read and dismounted again. If there are more than one requests for a tape, they all can be read in the same time.
The PROBLEM is to derive optimal visit orders of the tapes.
MODELING: We model this system as an unlimited-batch-service polling network, and use Hamiltonian cycle approach for the movement of the server.
Polling systems are networks of several queues attanded by a single server. Many real-world systems in manufacturing, communications, maintenance, road traffic control, data management and other areas can be formulated as polling systems.
Analysis: We analyze three service regimes: Globally Gated,
Locally Gated and Exhaustive, and consider three performance measures: 1. Minimizing weighted expected sojourn times of all jobs within a cycle. 2. Minimizing next cycle duration, and 3. Maximizing weighted throughput during a cycle.
For each combination of performance measure and gating procedure we derive optimal dynamic visit order policies. Some of the problems turn out to be combinatorially hard, but others yield elagant, easy to implement, index-type rules. We will give intuitive explanations to some of the results. This ia a joint work with Jan.v.d. Wal from Eindhoven University.
IMPORTANT: No previous knowledge of Polling Systems is required.
|
|
|
|
|
|
|
|