Professional Documents
Culture Documents
PICST - 2016 - A Network Load Balancing Algorithm
PICST - 2016 - A Network Load Balancing Algorithm
PICST - 2016 - A Network Load Balancing Algorithm
Abstract — Overlay solutions of SDN-based networks the physical network as needed, without changing
became very popular. Controller as a centralized anything on the virtual level.
management devices gather information about
The analysis of existing solutions [5, 6] and
network topology and current network state. Overlay
specification shows that for today the SDN concept have
network have a lot benefits however the using of
not strict requirements of load balancing and traffic
classical traffic engineering principles for traffic
distribution [7]. The lack of traffic engineering
management in Software-Defined Network is becoming
mechanisms such as load balancing algorithms, task
problem. The physical network does not automatically
scheduling and routing mechanisms raises the problem
adapt to the changes. As the overlay-based networks
connected with performance of overlay network solutions.
grow and requires more resources. The management
mechanism - network load balancing algorithm - that
take into account centralized structure and II. LOAD BALANCING ALGORITHM FOR SOFTWARE-
heterogeneous traffic nature of Software-Defined DEFINED NETWORK
Network is proposed in the paper. The main task that need to be solved for provision high
performance and efficiency of Software-Defined Network
Keywords— Software-Defined Networking; overlay; load infrastructure is optimal distribution of network and
balancing algorithm; adjacency matrix computing resources. The solution of this task should take
into account the optimal distribution such network
characteristics as channels throughput and compute node
I. INTRODCREATEUCTION
performance.
The main goal of Software-Defined Networking
Formally, the task of optimal distributing the channels
(SDN) concept is simplify the functionality and services
throughput can be represented as follow:
delivery due to decoupling network infrastructure into two
logical layers: data plane and control plan. The data Q( IN ,{Th ini ...Th inj }) → max , (1)
CP
forwarding and services delivery is totally managed by
control layer [1, 2]. Many SDN architecture solutions are
where Q() is optimization function, IN is total input
based on principles of network overlay. Implementation of
SDN as an overlay network is based on virtual switches {Thini ...Thinj}
data flow, is channels throughput, СP is
and tunneling protocols (VXLAN, NVGRE, etc.). The
basic idea is network overlay due to using tuned software policy of traffic distribution. The end goal of traffic
switches on the servers, where the virtual machine are distribution task can be represent as:
running [3, 4]. IN → max | Th ini ...Th outj . (2)
CP
Implementation of SDN as an overlay network allows
separate virtual network topology and configuration from The formalization (2) is achieved by the primary
the physical network topology. The virtual network creates choice of channel with a maximum throughput for highest
the path and forward data between virtual components, priority data flow and ranging the rest channel according
physical network only delivers packets to destination next principle: less traffic priority to the less throughput.
node. This approach allows rebuilding and modernizing The Edmonds-Karp algorithm [8] and maximum
throughput scheduling algorithm [9] are laying in the basis 0, p inp12 , p inp13 ,... p inp1M
of proposed load balancing algorithm. The proposed p
, 0, pinp23 ,... pinp2M
algorithm include next steps: M p inp = inp21
.........................
1. Controller executes topology discovery and p
inpM1 , inpM2 , inpM3 ,...0
p p
3. The construction of “adjacency matrix” are The total flow of incoming traffic is defined as
corresponding to the rule: if the link between nodes exist inpij = pinpij + npinpij , where p inpij is high priority
the interception between them match as “1”, if the link is
absent the interception symbol is “0”. Only those channels incoming data, npinpij is non-priority incoming data.
of communication are considered in the future work of the
algorithm, the value of which in the matrix is equal “1”, 5. Traffic distribution between nodes N i and N j
Chij = 1 . according to the weight of link in overlay network.
Firstly, the channel with maximal available throughput
4. Weight Determination and scheduling of traffic between nodes N i and N j is defined. Both direct channel
distributing. Two additional matrices are built on basic of
primary adjacency matrix M. and bypass channel can be chosen. Then the ratio of
incoming data flow intensity to throughput is evaluated.
First matrix M1 interprets the current throughput of The conclusion about possibility to use the channels for
network channels: if the link between nodes N i and N j priority data flow bases on next formalisms:
exists the meaning Chij = 1 changes to the current value of Ch iсij
1, Ch сij is overloaded
current throughput of channels. inpij
Ch iсij
0, c12 , c13 ,...c1M 1, Ch сij is oloaded , (3)
inpij
c , 0, c13 ,...c 2M ,
M1 = 21 Ch iс
.........................
ij
Th сij , underloaded
inpij
c M1 , c M 2 , c M3 ,...0
where Th сij is current threshold value for possible
The interception of rows and columns of matrix M 2
interprets the maximum bandwidth of channel for overlay channel load.
network. 6. The calculation of the residual channel resource.
0, C12 , C13 ,...C1M The calculation of residual bandwidth of chosen
channel bases on the collected statistics according to the
C , 0, C13 ,...C2M ,
M 2 = 21 next rules:
.........................
CM1 , C M 2 , C M3 ,...0
n
PIC S&T`2016
2016 Third International Scientific-Practical Conference
Problems of Infocommunications. Science and Technology