10-06-2017, 07:13 AM
An Algorithm for Dynamic Load Balancing in
Distributed Systems with Multiple Supporting
Nodes by Exploiting the Interrupt Service
INTRODUCTION
Rapid growth in use of computer has increased the
number of resource sharing application which
increased the amount of load across internet. The Problem can
be solved by increasing the size of servers or distributing
the applications in effective manner across different
servers which is termed as load balancing.load estimation can be estimated by means of processing power of the node.
Processing power means not only the processing
speed of Processorbut also the overall configuration
of node. In case of static algorithms, they collect no information and make probabilistic
balancing decisions, while dynamic algorithms collect
varying amounts of state information to make their
decisions.
It is its cost that limits the dynamic
algorithms, but at the high end of complexity are the
dynamic algorithms which collects varying amounts of
information.length. Load
balancing is done
to reduce mean job response time
under job transfer overhead ; and to
increase the performance of
each host.
Also, Small jobs will not suffer from starvation.
The commonly recognized categories of load balancing algorithms are 1)source-initiative algorithms, in which the
hosts where jobs arrive and take the initiative to transfer
the jobs, and:, 2) server-initiative algorithms, hosts
able and willing to receive transferred jobs go out to find
such jobs.Dynamic load balancing is complex but the
benefits form dynamic approach is much more than its
complexity.
PRIMARY APPROACH FOR DYNAMIC LOAD
BALANCING
A distributed system consists of independent
workstations which are connected usually by a local area network
Static load balancing don t fulfill the requirements for
load balancing. As in static load balancing, number of
jobs at a station is fixed. Dynamic load balancing does
the process while job are in execution. Jobs are allocated
to host or node. Processes are migrated from heavily
loaded node to light weighted node.
3. CENTRALIZED APPROACH FOR LOAD
BALANCING
If a heavily loaded node doesn t find
node in its cluster and due to congestion in network, node
fail to search the node far away cluster.
it is better that if heavily loaded node finds a temporary node in same
cluster to handle the over load. So, in centralized
approach one centralized node is provided in each cluster.The overload from nodes is transferred to
centralized node to increase output of each node.
4. MODIFIED APPROACH FOR DYNAMIC LOAD
BALANCING
In Centralized approach there is single node, so it processes
the load at high speed by using switching but still a
limitation exists. Away to remove the
limitation is to split the centralized node into small nodes
called supporting nodes (SNs). But still here supporting
node are not allotted load initially. Many times
supporting nodes is idle or they are not properly loaded as
only overload is assigned to supporting nodes. This is
wastage of power of supporting nodes. We can also use
the free time of SN by making them busy for this free
time.
5. ALGORITHM FOR MODIFIED APPROACH
There aretwo types of nodes. They are Primary and supporting
nodes. Primary nodes the are
main nodes and supporting are used to handle overload .Primary node tries to approach supporting node and
will find suitable supporting node, after finding suitable
and interrupts SN for execution of its process.
conclusion
for dynamics of a
distributed computing system in the context of load
balancing,
a Modified Model has been formulated. the centralized model was used for
solving the purpose of load balancing initially.
The aim in distributed system is to execute the
process at minimum cost i.e. time is most important
factor can be considered in cost calculation.
New dynamic load balancing
policy achieves a higher success,when compared to the
previously used load balancing techniques.
Ant colony optimization is used to minimize the
complexity for the purpose.