|
|
|
|
|
|
|
It is known that for DHR (Decreasing Hazard Rate) service times the FB (Foreground-Background) discipline minimizes the mean delay in the M/G/1 queue among all work-conserving and non-anticipating service disciplines. It is also believed that a similar result is valid for IMRL (Increasing Mean Residual Lifetime) service times. However, we point out a ï¬,aw in an earlier proof of this latter result and construct a counter-example that demonstrates that FB is not necessarily optimal within class IMRL. We will conclude the talk discussing the application of the Gittins index policy to the problem.
This is a joint work with Samuli Aalto (TKK)
|
|
|
|
|
|
|
|