An Approach to Auction-Based Web Server Admission Control - Developing Concepts in Applied Intelligence - page 103

The third assumption is that every class has a specific bandwidth, i.e.:

- bronze class (128kb/s),
- silver class (256 kb/s),
- gold class (512 kb/s).

Phase 2. The second stage of this algorithm takes a profitability metrics into
consideration. For every offer an index will be calculated and then the offers with the
highest value will be scheduled until the bandwidth is given. The profitability metrics
φ n can be defined as follows (1):

p

u k

ϕ =

(1)

i

n

t

i

where

sizeof ( r )

j

t =

(2)

n

b

u k

i

u i b (2).
The classification to some classes in the second stage doesn't matter.
During this stage following set of offers will be chosen (3):

t n - time of transfer j- th resource for bandwidth

k

L

∑
=

S = { o } => max ϕ

(3)

d n n

i 1

and by assumption (4)

L

∑
=

b ≤ 40 % B

(4)

u i k

i 1

Phase 3 . The third phase is used to serve the dynamic requests. In case, when during
the defined scheduling interval T some requests are rejected, the free bandwidth will
be assigned to dynamic offers. The price will be calculated basing on the individual
offer and requested bandwidth. The assigned bandwidth will base also on separately
negotiation process. The third phase takes place only during the scheduling interval
(when 0 < dynamic scheduling time < T ).

4.2 Profitability-Based Auction Algorithm

This approach is based only on the value of profitability index defined in the second stage
of the class-based auction algorithm. The following set of offers will be chosen (5):

L

∑
=

S = { o } => max ϕ

(5)

d n n

i 1