Many researches showed the ability of advance reservation to improve the predictability of the
system; that allows it to deliver the applications required time constrains. Applications with
many tasks require the system to ensure a number of reserv
ations on many different distributed
resources, which usually carried out through multi-level negotiation adding by that additional
overhead on the application total response time. The extra overhead depends on many
parameters including system workload and contention. Workflows add more complexity due to
their tasks’ dependencies; thus, any rejection of or delay for a task reservation would increase
application complete time.
This paper suggests the use of elastic advanced reservation plans that depend on time gaps
presented in the sub-optimal schedules, in order to improve the reservation acceptance rate. It
presents an elastic co-reservation agent which provides the needed reservations using First Fit
allocation strategy.
The results show the ability of the proposed agent to always improve the acceptance rate with an
average of (22.25%). The more important came out result is the agent ability to increase the
reservation acceptance rate with the increasing of system competence, reaching (48.4%) for
simultaneous 90 users at the system.