CloudDB AutoAdmin - Large Scale and Big Data: Processing and Management - page 362

and to reason about a particular application's performance because the per-
formance of an application hosted on a virtual machine becomes a func-
tion of applications running in other virtual machines hosted on the same
physical machine. In addition, it may be challenging to harness the full per-
formance of the underlying hardware, given the additional layers of indirec-
tion in virtualized resource management [14].
In practice, it is a very challenging goal to delegate the management of the SLA
requirements of the consumer applications to the side of the cloud service provider
due to the wide heterogeneity in the workload characteristics, details, and granular-
ity of SLA requirements, and cost management objectives of the very large number
of consumer applications (tenants) that can be simultaneously running in a cloud
environment. Therefore, it becomes a significant issue for the cloud consumers to
be able to monitor and adjust the deployment of their systems if they intend to offer
viable service level agreements (SLAs) to their customers (end users) [17]. Failing to
achieve these goals will jeopardize the sustainable growth of cloud computing in the
future and may result in valuable applications being moved away from the cloud. In
the following sections, we present our consumer-centric approach for managing the
SLA requirements of cloud-hosted databases.
11.4 FRAMEWORK ARCHITECTURE
In this section, we present an overview of our consumer-centric framework that
enables the cloud consumer applications to declaratively define and manage their
SLA for the cloudhosted database tiers in terms of goals that are subjected to a num-
ber of constraints that are specific to their application requirements. The framework
also enables the consumer applications to declaratively define a set of application-
specific rules ( action rules ) where the admission control component of the database
tier needs to take corresponding actions to meet the expected system performance
or to reduce the cost of the allocated cloud resources when they are not efficiently
utilized. The framework continuously monitors the database workload, tracks the
satisfaction of the application-defined SLA, evaluates the condition of the action
rules, and takes the necessary actions when required. The design principles of our
framework architecture are to be application-independent and to require no code
modification on the consumer software applications that the framework will support.
In addition, the framework is database platform-agnostic and relies on virtualization-
based database replication mechanism. To achieve these goals, we rely on a data-
base proxying mechanism that provides the ability to forward database requests to
the underlying databases using an intermediate piece of software, the proxy, and to
return the results from those request transparently to the client program without the
need of having any database drivers installed. In particular, a database proxy soft-
ware is a simple program that sits between the client application and the database
server that can monitor, analyze, or transform their communications. Such flexibility
allows for a wide variety of uses such as load balancing, query analysis, and query
filtering.