Distributed Controller Placement in Software-Defined Networks with Consistency and Interoperability Problems

Table 8

Comparison of controller placement with load balance awareness.

Ref	Approach		Load balance optimization objectives	Problem formulation	Solution approach	Weakness
	CCA	SMA
[52]	X	√	Response time, throughput, migration time	NA	4-phase migration protocol	Not support HetCP, CMF, controller location, scalability
[112]	√	X	Latency, no of controllers	ILP	CPlex solver,	High overhead at master controller, CMF
[113]	√	X	Setup time	None	Flow request partition	CMF, additional overhead at CP
[114]	√	X	Flow setup time, comm overhead	ILP	Heuristic	Master controller overhead
[115]	√	X	Overhead	QIP	Greedy	No support HetCP, CMF
[116]	√	X	Setup time	None	Greedy	CMF, imprecise controllers’ load collection. Vulnerable
[117]	√	X	Latency, availability	Graph theory’s centrality stress	Heuristics, lattice graph	Not support HetCP, CMF
[118]	√	X	Response time	NA	Cluster vector	High overhead at master, not support HetCP, CMF
[119]	X	√	Flow setup time, utilization	Markov chain	Distributed hopping algorithm	Not support HetCP, CMF
[120]	X	√	Setup time	Bin-packing	Greedy algorithm	Not support HetCP, CMF
[121]	X	√	Throughput, load oscillation	Load informing strategy	Inhibition algorithm	Heterogeneous interoperability problem, CMF
[122]	X	√	Latency, saturation attack	3-D Earth mover model	Heuristics	Controller location, not support HetCP, CMF
[123]	X	√	Effective controller load	LP, graph theory	Heuristic	Not support HetCP, CMF
[124]	X	√	Setup time, overhead, latency	ILP	TOPSIS, heuristic	Not support HetCP, CMF
[125]	X	√	Response time, throughput	Game theory	Game theoretic	Not tested on a production network, does not support HetCP, CMF
[126]	X	√	Overhead, response time	NA	Heuristic	Not support HetCP,CMF
[127]	X	√	Failover, throughput, loss	Mathematical	Heuristic	Prolong end-to-end RTT, not support HetCP, CMF
[128]	X	√	Response time, throughput	Mathematical	Heuristics	Not support HetCP, CMF