12 S.S. Manvi, G.

Krishna Shyam / Journal of Network and Computer Applications ∎ (∎∎∎∎) ∎∎∎–∎∎∎

Table 8
Resource mapping schemes.

Name of the scheme Functioning

Mapping logical plane to underlying physical plane Presented a novel set of feasibility checks for node assignments based on graph cuts
(Hou et al., 2009)
Symmetric mapping pattern Presents the symmetric mapping pattern, an architectural pattern for the design of resource supply systems. It
(Grehant and Demeure, 2011) divides resource supply in three functions: (1) users and providers match and engage in resource supply
agreements, (2) users place tasks on subscribed resource containers, and (3) providers place supplied resource
containers on physical resources
Load-aware mapping (Chen et al., 2009) Explores how to simplify VM image management and reduce image preparation overhead by the multicast file
transferring and image caching/reusing. Additionally, the Load-Aware Mapping, a novel resource mapping
strategy, is proposed in order to further reduce deploying overhead and make efficient use of resources
Minimum congestion mapping (Bansal et al., 2011) Proposes a general framework for solving a natural graph mapping problem arising in cloud computing. And
applying this framework to obtain offline and online approximation algorithms for workloads given by depth-d
trees and complete graphs
Iterated local search based request partitioning A novel request partitioning approach based on iterated local search is introduced that facilitates the cost-
(Leivadeas et al., 2011) efficient and on-line splitting of user requests among eligible Cloud Service Providers (CSPs) within a networked
cloud environment
SOA API (Xabriel et al., 2012) The solution is designed to accept different resource usage prediction models and map QoS constraints to
resources from various IaaS providers
Impatient task mapping (Mehdi et al., 2011) Proposes batch mapping via genetic algorithms with throughput as a fitness function that can be used to map
jobs to cloud resources
Distributed ensembles of virtual appliances (DEVAs) Requirements are inferred by observing the behavior of the system under different conditions and creating a
(Villegas and Sadjadi, 2011) model that can be later used to obtain approximate parameters to provide the resources. These models are
usually measured by treating the application as a black-box (i.e., without employing any knowledge of the
internal implementation or design)
Opportunistic resource Adopts a simple greedy heuristic to all virtual nodes to sort in a decreasing order of their CPU constraints and
places them in a queue
Sharing and topology-aware node ranking (ORSTA) Then, maps each virtual node in the sorted queue to the unused substrate node with the highest rank
(Zhang et al., 2012)
Hence, minimizes the length of the substrate paths that virtual links are mapped to
Mapping a virtual network onto a substrate network Developed an effective method (using backbone mapping) for computing high quality mappings of virtual
(Lu and Turner (2006)) networks onto substrate networks. The computed virtual networks are constructed to have sufficient capacity to
accommodate any traffic pattern allowed by user-specified traffic constraints

5.2.1. Open challenges in resource allocation to maximize cloud utilization in IaaS by calculating the capacity of
The challenges in resource allocation are as follows. application requirements so that minimal cloud computing infra-
structure devices shall be procured and maintained. This can be
 How to design a resource allocation scheme that spans several achieved by using cognitive architecture that automatically builds
clusters and data centers? a model of the machine behavior based on prior training data.
 How to devise a mechanism that allows controlling the trade- In a cloud computing environment, a logical network (i.e a set
off between the cost of reconfiguration and maximizing the of virtual machines) must be deployed on to physical network
cloud utility? (servers). This requires mapping of VMs to physical resources.
 How to develop a tree-based protocol for resource manage- The mapping problem is dealt in Hou et al. (2009) which translates
ment in cloud environments and how such a protocol compares virtual machines assignment onto physical servers and assigns
with a gossip-based protocol with similar functionality? flows in the physical network with bandwidth allocation so that
 How to bring out the techniques for allocation of services to requirements of logical communication can be met.
applications depending on energy efficiency and expenditure of An allocation which is directed by a decision system under user
service providers? control can result in high resource supply costs and an allocation
 How and when to reallocate VMs to minimize the power drawn directed by a decision system under provider's control can result
by the cooling system, while preserving a safe temperature of in low user-perceived resource value. Instead of compromising
the resources and minimizing the migration overhead and with them, symmetric mapping referred in Grehant and Demeure
performance degradation? (2011) builds on these differences from the system design. It relies
 How to design SLA-oriented resource allocation strategies that on the idea that a system benefits from the involvement of
encompass customer-driven service management, computa- different participants if it induces them to adopt predictable
tional risk management, and autonomic management of clouds behaviors and uses these behaviors as part of its mechanism.
in order to improve the system efficiency, minimize violation of Chen et al. (2009) have worked towards an efficient resource
SLAs, and improve profitability of service providers. management system for on-line virtual cluster provision.
 How to move from one cloud to another cloud considering In particular, they focus on two crucial problems namely efficient
vendor lock-in issues? What if a good part of our application VM image management and intelligent resource mapping. Addi-
infrastructure resides with a single cloud provider? tionally, they have proposed an intelligent resource mapping
strategy, named load-aware mapping, in order to reduce deploying
overhead and balance resource utilization.
5.3. Resource mapping In cloud computing, the underlying resource is a physical
network (also called the substrate) consisting of servers that are
Mapping of virtual resources to physical resources has an inter-connected via communication links. The allocation of a
impact on cloud clients. Resource mapping is a system-building workload to the substrate can be viewed as mapping one graph
process that enables a community to identify existing resources into another. This consists of two aspects: (a) node-mapping, the
and match those resources to a specific purpose. The issue here is assignment of processes to servers, and (b) path-mapping, the

Please cite this article as: Manvi SS, Krishna Shyam G. Resource management for Infrastructure as a Service (IaaS) in cloud computing:
A survey. Journal of Network and Computer Applications (2013), http://dx.doi.org/10.1016/j.jnca.2013.10.004i