Cross-Layer Monitoring in Cloud Computing

Introduction

Traditionally, businesses had to invest in the acquisition of their own dedicated IT infrastructure to satisfy their computational requirements. This led to a planning problem, as requirements for IT infrastructure in businesses are constantly changing according to current workload and business strategy. As a consequence, traditional businesses have to make a significant up-front investment in IT infrastructure to provide enough computational power and storage capacity to handle the highest peaks of workload and successfully face future challenges.

Cloud Computing (Armbrust, et al., 2010) is an upcoming computing technology with the potential to change how businesses implement their services. The idea behind Cloud Computing is to use virtualized resources which are provided by a third-party. These resources are acquired on-demand and typically paid for in a pay-per-usage model (Buyyaa, Shin Yeo, Venugopal, Broberg, & Brandic, 2009). This paradigm enables Cloud customers to adapt the allocated IT infrastructure dynamically, based on changing business objectives and user demands. This leads to a significant reduction in investment needed to acquire IT infrastructure and creates a paradigm, which leads to a balance between own dedicated IT computation resources and IT resources rented from third parties for dealing with peaks of workload.

Cloud Computing provides services on a stack composed of three service layers (Hurwitz, Bloor, Kaufman, & Halper, 2009): Infrastructure as a Service (IaaS), Platform as a Service (PaaS), and Software as a Service (SaaS).

The IaaS layer exposes services for which the users can deploy virtual resources (e.g. virtual machines) on demand. This IaaS layer uses the underlying physical resources (e.g. physical machines, network, and storage components) to achieve the functionality. The PaaS layer relies on these virtualized resources to provide more complex middleware services. These middleware services are intended for being used as components and functionalities for other applications and services. Finally, the SaaS layer provides consumable high-level services to users (i.e. entire enterprise applications or web services). To provide these services, the SaaS layer uses the other layers of the cloud stack. Each layer available in the cloud stack is not restricted to end users but target different groups of users, i.e. developers and operations staff. [REMOVED REF FIELD] illustrates the types of services provided by the IaaS, PaaS, and SaaS layers and their primary user groups.

Each service layer may be provided by a different entity (service provider). Likewise, there may be different consumers for Cloud services (service consumer). In this context, a private Cloud is a usage scenario in which service provider and consumer are of the same party, whereas a public Cloud refers to a scenario in which provider and consumer are different parties and have to be from different administrative domains. Note that a private cloud can be considered as a software stack for internally governing the computational resources available in a company whereas a public cloud involves different providers of services.

Currently, almost all research works are focused on public Cloud scenarios, which are the predominant option in Cloud Computing due to imminent advantages for the service consumer, such as, a reduction of acquisition cost in IT infrastructure and outsourcing of IT maintenance tasks to the service provider. This book chapter is focused on a public Cloud scenario in which a service provider offers Cloud services, generally by means of the Internet, to a variable number of service consumers.

The design of a stack for a public Cloud scenario is generally challenging as the architecture has to reflect the individual requirements of different service consumers and service providers. Many challenges associated to the design of the service layers are still open issues nowadays and they are regarded as drawbacks for which some businesses are reluctant to move to the Cloud. One of the main concerns reported by Cloud customers (Fujitsu, 2010) is how consumers get full control over virtualized resources rented on-demand.