Using Multiple Data Stores in the Cloud: Challenges and Solutions - Data Management in Cloud, Grid and P2P Systems - page 87

Using Multiple Data Stores in the Cloud:
Challenges and Solutions
Rami Sellami and Bruno Defude
Institut TELECOM, CNRS UMR Samovar, Evry, France
{ rami.sellami,bruno.defude } @telecom-sudparis.eu
Abstract. Cloud computing has recently emerged as a new computing
paradigm. This latter provides provisioning of dynamically scalable and
often virtualized resources which are offered as services. In this context,
there is a services pool that supports platforms and mechanisms in order
to guarantee the applications development and execution. This set of
services is called platform as a service (PaaS). One of the main goals of
the PaaS is to support large data stores by ensuring elasticity, scalability
and portability. Many applications have to manage different types of
data that a single store can not eciently support. Consequently, clouds
need to deploy multiple data stores, allowing applications to choose those
corresponding to their data requirements. In this paper, we present and
discuss the requirements of such environments and analyze current state
of the art.
Keywords: Large data store, Cloud computing, Polyglot persistence,
Data management contract, PaaS, Data mediation, Data consistency.
1 Introduction
Cloud computing, a relatively recent term, has become nowadays a buzzword in
the Web applications world. Despite the importance of this paradigm, there is
no consensus on the cloud computing definition. In this context, experts present
a set of twenty one definitions of cloud computing [1]. Based on these defini-
tions, we can define cloud computing as a large scale distributed computing
paradigm based on virtualized computing and storage resources, and modern
Web technologies. Over the internet network, cloud computing provides scal-
able and abstracted resources as services. All these services are on demand and
offered on a pay-per-use model.
Cloud computing is often presented at three levels: Infrastructure as a Service
(IaaS) where clients can deploy their own software, Platform-as-a-Service (PaaS)
where clients can program their own applications using the services of the PaaS
and finally Sofwtare as a Service (SaaS) where clients use existing applications
of the cloud. In this paper, we focus on the PaaS level. Several PaaS commercial
solutions exist (e.g. Microsoft Azure [2], Salesforce Force.com [3], Google App
Engine [4], etc.) and many research projects are seeking to improve the security,
scalability, elasticity and interoperability in the PaaS.