See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/270791476

Cloud Database Management System Architecture

Article · January 2013

CITATIONS READS
38 17,896

2 authors:

Mansaf Alam Kashish Ara Shakil

Jamia Millia Islamia Princess Nora bint Abdul Rahman University
133 PUBLICATIONS   1,041 CITATIONS    41 PUBLICATIONS   518 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Resource Allocation in Cloud Environment View project

study of workloads View project

All content following this page was uploaded by Mansaf Alam on 20 February 2015.

The user has requested enhancement of the downloaded file.

 UACEE International Journal of Computer Science and its Applications - Volume 3: Issue 1 [ISSN 2250 - 3765]

Cloud Database Management System Architecture

Mansaf Alam and Kashish Ara Shakil

Abstract— One of the major challenges for the IT companies Cloud computing is a very promising technology for the
today is how to manage large growing volumes of data and how to future. Cloud computing can cause a major cut down in the
produce a quality driven software product ensuring optimal marketing time by taking charge of provisioning of resources
utilization of resources with minimum cost. The database such as servers, hardware or any other computing resources. It
management system is a software system i.e. a set of programs that
can lead to cost reductions as it employs the use of pay per use
provides its users with processes for defining, building,
manipulating and sharing databases amongst the users and and also provides a better utilization of resources. Thus, cloud
applications. A cloud data base management system is a database Computing is also forecasted as a Green computing
management system for management of cloud data and provides technology. Cloud computing also promises (virtually) an
delivery of computing as services rather than as product. In this infinite scalability along with flexibility [3].The probable
paper we have proposed an architecture for management of data in benefits of cloud computing are astounding. Therefore, in
cloud termed as “Cloud Database Management System order to attain these benefits it is required that each aspect of
Architecture”. The cloud database management system provides an cloud platform must support the key design principles of cloud
approach for management of cloud data. The cloud data are spread model. Some of the key design principles are dynamic
over the internet and are stored to a remote server managed by a
scalability, availability, ability to allocate and reallocate
third party. Hence, the cloud data management is a major issue
which needs to be catered to. A well-defined architecture is thus resources. But, a majority of database servers are not able to
required to manage the cloud data, available at a remote location. meet these requisites [11]. Hence, requirements for a database
In this work an architectural model for cloud database in cloud are very different from traditional databases, as the
management system has been developed .This architecture is based cloud environment is very unpredictable [9] .Every database in
on the three schema architecture for data base management system cloud must be highly available and reliable. Attainment of
and three level object oriented database management system scalability is also a very complex process. Also, it is very
architecture. difficult to maintain distributed multiple copies of database at
different locations. Thus database in cloud needs to be
Keywords— Three Schema architecture, cloud system
architecture, green computing, three level object
accessed and managed in a distinguished manner i.e. a proper
oriented database architecture framework for access and management of cloud data are
essential.
In this paper we have proposed an architecture for cloud
. Introduction
I
database management which aims at addressing these
requirements of cloud databases.
The major concern for any organization today is how to
manage the ever increasing huge volumes of data and to
deliver more and more enhanced services with reduced cost.
II. Related Work
The need of the hour is management of these huge volumes of Our proposed approach “Cloud database management system
data along with ensuring scalability, availability and reliability architecture”, bears a close resemblance to the three schema
.This concern is what acted as the stepping stone for architecture for database management [8] and Three Level
development of cloud. Object Oriented Database Architecture Based on Virtual
Cloud computing enables IT resources such as the developers Updatable Views [7].
to concentrate more on the core issues like development of
product, rather than worrying about secondary issues like A. Three Schema Architecture
availability of servers, storage space etc. It enables cloud users The Three Schema architecture is a representative architecture
to use unlimited computing powers by renting more and more of database systems. It provides support for multiple user
resources via cloud computing. views and program data independence with the aim to separate
the user applications and physical database. In this architecture
Mansaf Alam three levels are defined Fig. 1 [6]
Jamia Millia Islamia 1) Internal level: this level is composed of the internal schema,
India which describes the actual physical storage of data.

2) Conceptual level: this level is composed of the conceptual

Kashish Ara Shakil
schema and describes the structure of whole database for a
Jamia Millia Islamia community of users. It hides details of data storage at physical
India level and concentrates on describing entities, relationships, user

27
 UACEE International Journal of Computer Science and its Applications - Volume 3: Issue 1 [ISSN 2250 - 3765]

Thus, the above three participants work together to provide

management of user privileges along with directing database
application development. Hence the Three level object
oriented database architecture is well suited for object oriented
database.

Figure 1. Three Schema architecture

operations and other constraints.

3) External level: This level provides an external view to end
users i.e. it provides the end users with that part of
database in which they are interested and hides other low
level details
The above three schema architecture is well suited to the needs
of a relational database environment but not for a cloud
environment as a cloud environment requires access by many Figure 2.A conceptual view of three levels Object Oriented Architecture
kinds of users having different service requirements therefore,
it requires a greater level of customization. Also cloud Since we know that the computing environment for a cloud
environments have stringent security and privacy requirements based system is very different from traditional database
which cannot be fulfilled with this architecture. environments and for object-oriented and XML-oriented
environments with emphasis on user data privacy, security,
scalability elasticity, availability of resources etc. Different
B. Three-Level Object-Oriented Database cloud clients may need different types of customized services
Architecture and access controls as per their needs and requirements. Thus,
Three level object oriented database architecture [7] is in order to overcome the limitations of applicability of three
architecture for object oriented database access and schema architecture for database and Three-Level Object-
management, is based on updatable views which provides a Oriented Database Architecture we have proposed an
mapping of stored objects onto virtual objects. In this architecture for cloud data access which has physical, logical
architecture the middle layer is referred as a DBMS managed and technical advantages over the existing architecture for
middle layer. Its features are that it is transparent, provides database access.
users with the ease of management and modifications .This Following are some of the major issues catered in our
architecture defines the following user roles and proposed approach:
responsibilities.Fig.2  Customization: Since cloud data are to be used by a
Database programmer is responsible for creating internal and variety of different clients with different resource
conceptual schema of the data based upon previously created requirements. Each user of a cloud will have
design, as per the business requirements. resources available via cloud as per their
requirements.
Database administrator is responsible for defining external  Conceptual modeling: As a cloud data will be used by
schema for particular users. She/he creates updateable views clients with different types of requirements.
which are built upon data store. Therefore each user will have a different view of
cloud data.
Application programmer is a database user, who uses the  Information hiding: A cloud user should be
database and is well versed with interfaces of views provided constrained to use only the part of service which
by the database administrator. he/she has requested.

28
 UACEE International Journal of Computer Science and its Applications - Volume 3: Issue 1 [ISSN 2250 - 3765]

 Security, privacy and cloud data access: cloud data it’s very important to ensure that data consistency, scalability
base administrators/servers should have flexible and security are maintained. In order to address these issues
facilities to grant and access resources to the users and several other critical issues regarding data, a data base
according to the access rights granted to them. management system for cloud data is imperative.
 Orchestration/choreography of services: it’s In cloud two primary DBMS architectures are used shared
extremely important in cloud computing as it is based nothing and shared disk. Shared nothing is a distributed
on scaling and pay per use, connecting and computing architecture in which each node is self-sufficient
automating of work flows and hence cloud services and is independent of any other node i.e. each node in shared
must be orchestrated at every layer. nothing architecture has its own memory and disk storage and
does not share it with any other node. There is no point of
. Cloud Users
III
contention in the nodes.
Shared disk architecture is a computing architecture in which
Cloud users play a very vital role in a cloud computing each node has its own memory but they share disk storage
application. It’s the cloud users whose data are ultimately space. It actually partitions the data such that each database
managed by the cloud service providers. Mladen A. Vouk[4] server processes and maintains its own piece of database [5].
has given a hierarchy of the different cloud users .The cloud In this paper we will first discuss shared nothing and shared
user hierarchy consists of four users: disk architectures, their limitations and then propose our cloud
A. Cloud Infrastructure developers: data base management system architecture.
These are developers who are experts in specialized areas
such as networks, computational hardware, storage, A. Shared nothing
operating system imaging etc and are responsible for In shared nothing architecture each node has a memory as well
development, maintenance and administration of cloud as storage space of its own. These nodes communicate to each
framework. They are also responsible for hiding lower other by message passing through an interconnected network.
level details from its clients In a clustered system only one resource can be accessed at a
time ,during failure resource ownership can be transferred to
B. Service Authors: any other resource connected on the network[5].Shared
These are developers of base line images and services, nothing has a virtually infinite scaling capability as each of the
which might be used directly or integrated into other node is isolated from other. This approach is suitable for
services. This enables cloud users who want to use image applications with heavy data update requirements. Shared
creation tools and service management tools to nothing does not provide inherent data consistency like shared
concentrate on their development prospects rather than the disk as each node has its own set of memory therefore
details of cloud infrastructure. transactions must be configured in order to ensure consistency
of data. Since, in shared nothing architecture each node has its
C. Service integration and provisioning:
own memory it has problem of load balancing as load of one
server can’t be shared with other servers.
Service integration and provisioning experts are
responsible for creation of composite solutions required
by an end user. They create new customized services for B. Shared Disk
the clients by making updating the existing services and In shared disk architecture all the connected nodes share the
images. same storage disk space. It’s particularly suitable for
applications in which distributing or partitioning of workload
D. End Users:
is very difficult. Shared disk is based on message or token
passing each node communicates with every other node by
They are the most important users of cloud services .They passing messages.
are the users who ultimately use the cloud services Shared disk cannot scale up as well as shared nothing
provides by the cloud providers. They require that the architecture but provides an advantage of dynamic load
services made available to them must be reliable, secure, balancing which is the driving factor that accommodates
easy to use and scalable. temporal and evolutionary changes in usage patterns
[5].Another advantage of shared disk is that it maintains
IV. Cloud Database consistency of data as there is only one copy of data shared by
Database is an organized collection of data, and is the heart all the nodes.
and soul of any information system. Cloud infrastructure Therefore we can conclude that both shared nothing and
consists of huge volumes of data which might be shared shared disk architectures can be used in cloud depending upon
amongst multiple tenants .Thus, data management in particular the requirements of the organizations, developers and
is an essential aspect for storage in cloud. The data are customers. But it is necessary to have a separate architecture
distributed in cloud across multiple locations and might for cloud data base itself in order to overcome the limitations
contain certain privilege and authentic information. Therefore of shared disk and shared nothing architectures. We will now
discuss our proposed approach “Cloud database management

29
 UACEE International Journal of Computer Science and its Applications - Volume 3: Issue 1 [ISSN 2250 - 3765]

architecture”, which is based on three tier architecture of availability of cloud data at all times, provides facility of
DBMS and three level object oriented database architecture. multitenancy, on demand self-service, elasticity and various
other characteristics of a cloud. It adds the feature of data
V . Cloud Database Management abstraction in a cloud .It hides the details of data storage at the
datacenter level and makes the underlying software and
System Architecture database transparent to its client users. It provides customized
Now, we will discuss our proposed approach for cloud cloud facilities as per the customer’s requirements. It’s this
database access and management which is based on three part which makes decisions regarding access rights of users
schema architecture and three level object oriented database etc.
architecture [7] and the different roles at each of the levels. It The cloud service provider level can in turn be divided into
provides foundation for three level database architecture. two sublevels on the basis of functionalities at each of these
Cloud database management system architecture represents levels. This level is subdivided as administrative level and
data in three levels Data Center Level, Cloud Service provider security level:
level and Client Level: Fig. 3.
1) Administrative level
At this level database instances are controlled using a service
API. This API is available to the end users and enables
database users to scale and maintain their database instances.
For example, the Amazon Relational Database Service's
service API allows creation and deletion of database instances.
It also has equipped with the power of modifying resources
available to database instances, creating a backup of data and
also restoring the database instance in case of failure.
It is responsible for providing scalability of resources by
adding more resources as per the user requirements.
Scalability can either be provided automatically or users may
have the privilege to scale up or down as per their
requirements through an API.
High availability of database which is a commitment of cloud
service providers is also taken care of at this level.

2) Security level
This level is responsible for providing security control to the
cloud user’s data and to assure continuous correctness of their
stored data without any local copies. It ensures end to end data
Figure 3. Cloud database management system architecture confidentiality, prevents data loss. User authentication, data
encryption, intrusion detection are some of the measures
A. Data Center level adopted to ensure cloud security.
This level represents the actual physical storage location of
data in cloud and consists of several servers catering to the C. Client level
needs of cloud users Since we know in a cloud infrastructure
data are stored at the data center .This level is a representative This level consists of cloud users or client computers. It is the
of a cloud database. In cloud computing data are stored as a most visible level to the clouds end users[10].At this level
virtualized pool of storage. Cloud providers operate these data cloud users have the view of that part of cloud database that a
centers as per the requirements of the clients. The data center particular user is interested while all the other details of cloud
operators provide the cloud users with a virtualized illusion of service provider level and data center level remain hidden
the resources according to their requirements, and expose from the end users. At this level, data center operators provide
them with storage pools where the customers can store their the cloud users with a virtualized illusion of the resources as
files and data. The cloud algebra [12] is used to manipulate the per their requirements, and expose them with storage pools
data available in data center. where the customers can store their files and data.
End users of a cloud are client computers that access the cloud
data. Client of a cloud infrastructure is not necessarily a
B. Cloud Service provider level computer but it can also be any other computing device such
At this level building and management of cloud applications is as web browser, mobile apps, mobile phone, tablet etc. End
done. This is a middleware level and consists of several users make use of cloud computing infrastructure on the basis
distributed servers catering to the needs of cloud users. It is of pay per cost model thereby reducing the overall costs of
these servers that are responsible for providing all the facilities developing their applications.
promised by a cloud service provider to its users. It ensures the

30
 UACEE International Journal of Computer Science and its Applications - Volume 3: Issue 1 [ISSN 2250 - 3765]

. Roles In Cloud Database

VI VII . Conclusion
Management System Architecture Three Schema Architecture and Three-Level Object-Oriented
Database Architecture are two well defined architectures for
The main aim of a database management system is that they relational dbms and object oriented data respectively. In cloud,
provide a clear separation between a user application and data are distributed over the cloud across several distributed
actual physical data base along with ease of management and servers and cloud database has certain special requirements
modifications. Therefore in order to ensure a systematic access like scalability, availability along with stringent security and
of cloud data base for cloud users and for providing all the user authentication needs. As, these requirements cannot be
facilities of cloud database management system it is vital to fulfilled with the existing architectures for relational and
define clearly the roles at each level in our cloud database object oriented data thus we proposed our architecture for
management system along with the responsibilities .The cloud data called Cloud Database Management System
following are the roles and responsibilities: Architecture.
Our, architecture is based on ANSI/SPARC three schema
A. Cloud Application Programmer architecture and three level object oriented architecture. We
Cloud Application Programmer is the actual database user, have defined cloud database to be organized into a hierarchy
who uses the database as service provided to him or her by the of three levels cloud data center level, cloud service provider
cloud service provider and creates the database schema. The level and client level. We have also identified roles defined at
application programmer operates at the client level, accesses each level i.e. roles of Cloud Application Programmer, Cloud
the cloud databases via an API provide by the client service database administrator and Data center programmer. Several
provider as per his /her requirements. The programmer can related issues like how the security measures are adopted in
perform tasks on the cloud database similar to a tasks database and other issues like administration have not been
performed on traditional databases like data retrieval, fully voiced here and are potential topics of research.
manipulation, modification, deletion and storage. However the For our future research work we have planned to work on the
security of this data is handled by the database administrator. cloud service provider level and perform a deep study of
administrative and security levels, and aims at addressing the
B. Cloud Database Administrator issues with the existing clouds security and issues with
administration as well.
Cloud database administrator operates at the client service
provider level of a cloud database management system. They References
are the driving force behind the entire cloud database
management system as they are responsible for providing [1] Gerard Conway and Edward Curry, “Managing Cloud Computing: A life
database as a service to the cloud users .They provide cycle approach”, 2nd International Conference on Cloud Computing
customized external schemata for clients as per their and Services Science (CLOSER 2012), 2012, pp. 198-207.
requirements i.e. providing cloud users with the part of [2] Ashraf Aboulnaga, Kenneth Salem, Ahmed A. Soror, Umar Farooq
Minhas,Peter Kokosielis, Sunil Kamath “Deploying Database
database service they are interested in, hiding all the data Appliances in the Cloud”,Bulletin of the IEEE Computer Society
center level information. They are the security inspectors and Technical Committee on Data Engineering”, 2009.
in charge of data authentication, granting privileges and access [3] Donald Kossmann, Tim Kraska, Simon Loesing, “An Evaluation of
rights to users. Scalability and availability of cloud database Alternative Architectures for Transaction Processing in the Cloud”,
SIGMOD’10, 2010.
instances is also handled by the cloud database administrators.
[4] Mladen A. Vouk, “Cloud Computing – Issues ,Research and
Implementations", Journal of Computing and Information Technology -
C. Data Center Programmer CIT , 2008,pp 235–246.
Data center programmers/operators operate at the data center [5] Sunguk Lee, “Shared-Nothing vs. Shared-Disk Cloud Database
Architecture”, International Journal of Energy, Information and
level. They are in charge of storage of cloud data across Communications Vol. 2, Issue 4, 2011.
several servers and for ensuring integrity of data. They define [6] http://www.dcs.vein.hu/CIR/cikkek/Database_Environment.pdf
the behavior of the stored cloud data. Thus, our proposed well [7] Piotr Habela1, Krzysztof Stencel, Kazimierz Subieta, “Three-Level
defined cloud database management system architecture has Object- Oriented Database Architecture Based on Virtual Updateable
the following advantages over traditional DBMS: Views1” ADVIS 2006,Fourth Biennial International Conference on
Advances in Information Systems, Volume 4243, 2006, 2006,pp 80-89.
a) User control over provisioning and management of data. It
[8] D.C. Tsichritzis, A. Klug (eds.): The ANSI/X3/SPARC DBMS
enables users to manage and store data themselves. Framework: Report of the Study Group on Data Base Management
b) Helps in storage of files in different format’s as per the user Systems, Information Systems , 1978.
requirements like text files, image files, xml files etc. [9] Dave Rosenberg, “Are database in cloud really all that
c) It provides cloud users to perform all the tasks which can be different?,CNET” , 2011.
performed on a normal DBMS like data manipulation, data [10] Lamia Youseff, Maria Butrico, Dilma Da Silva, “Towards a Unified
storage, data retrieval etc. cloud Computing”, Grid Computing Environments Workshop, GCE
'08, 2008.
d) It is cheaper than the traditional dbms.
[11] http://www.scaledb.com/pdfs/CloudComputingDaaS.pdf
e) It is more accessible.
[12] Mansaf alam, “ Cloud algebra for cloud database management system“,
CCSEIT '12, ACM New York, NY, USA, 2012 pp 26-29.

31

View publication stats