VIDaaS

Design Architecture of Virtual Infrastructure with Database as a Service

Asif Akram and Christian Fernau

Oxford University Computing Services, Oxford University, Banbury Road, Oxford, U.K.

Keywords: Cloud, Cloud Computing, Software as a Service (SaaS), Database as a Service (DaaS).

Abstract: This paper presents design and deployment architecture of research project Virtual Infrastructure with

Database as a Service (VIDaaS). VIDaaS supports common requirements of multi-disciplinary researchers

and caters for different database types besides the traditional relational database. The ambitious and

challenging goals of VIDaaS require innovative and flexible design and deployment architecture. The

primary focus of this research paper is to discuss an architecture framework for the existing database system

melded with the Cloud computing paradigm in order to improve the scalability and boost the performance

of database systems. VIDaaS is based on reusable architectural model benefiting from Service-Oriented

Architecture (SOA) and cloud design which is not limited to DaaS and can be applicable to any Software as

a Service (SaaS).

1 INTRODUCTION

The Virtual Infrastructure with Database as a

Service (VIDaaS) Project comprises two

fundamental elements: the deployment of a

‘Database as a Service’ (DaaS) production-quality

software service; and the implementation of a hybrid

virtual infrastructure (VI) upon which DaaS, can be

developed, refined, run and managed. The hybrid

'Infrastructure as a Service' (IaaS) component

provides a virtual infrastructure upon which the

DaaS and other 'Software as a Service' components

can be hosted. This involves enabling seamless

capacity management and the sharing of

infrastructure and services between universities

and/or other academic institutions.

The DaaS component is developed from the

prototype delivered for the Supporting Data

Management Infrastructure for the Humanities

(Sudamih) Project to a full production service.

Functionality is extended to meet common

requirements of researchers beyond the humanities

and to cater for different database types besides the

traditional relational database. From the user

perspective, it offers an intuitive means of creating,

editing, querying and sharing databases through a

simple Web interface.

The VIDaaS Infrastructure as a Service (IaaS) is

secure, resilient, reliable, cost-effective and

necessarily include access management and

accounting tools. VIDaaS aims in delivering a

platform between researchers and the low level

infrastructure services found on the commercial

cloud, which will avoid users being locked in to any

one commercial cloud provider (Logicalis, Amazon ,

Salesforce ) and proprietary technologies (Amazon

SimpleDB , Microsoft SQL Azure , Amazon

Relational Database Service , DabbleDB, Google

BigTable). We reckon as a service provider of

outsourced databases, VIDaaS can leverage an

economy of scale by providing complex systems for

many users. Users, in turn, could save large amounts

of money on database experts, security solutions and

software and hardware maintenance. The VIDaaS is

developed from various open source java

frameworks and is compliant to Java Enterprise

Edition 6 (JEE6) specification.

2 SYSTEM ARCHITECTURE

VIDaaS consists of various reusable components

and each component is an independent fully

functional service. VIDaaS components deal with

Identity and Access management; Payment and

Billing; on demand Virtualization; and VIDaaS

controller. The top level architecture diagram is

shown in the figure 1.

102

Akram A. and Fernau C..

VIDaaS - Design Architecture of Virtual Infrastructure with Database as a Service.

DOI: 10.5220/0003932301020107

In Proceedings of the 2nd International Conference on Cloud Computing and Services Science (CLOSER-2012), pages 102-107

ISBN: 978-989-8565-05-1

 2012 SCITEPRESS (Science and Technology Publications, Lda.)

Figure 1: Top level VIDaaS deployment architecture.

The VIDaaS is based on “split application

functions and couple loosely”. The communication

between components is via secure RESTful Web

services. VIDaaS communication layer uses

network-based interfaces and discourages inter-

process communication or file based communication

paradigms. This allows VIDaaS to effectively scale

in the cloud, as demand grows, the system can scale

each component independently instead of having to

scale the entire application. The separation and

reusability of functionality inherent in SOA make it

an ideal architecture for the cloud. Each of the

VIDaaS components is discussed below with the

role it plays in the whole system.

2.1 VIDaaS Admin Node

The “VIDaaS Admin Node” glues together all

VIDaaS components together in compliance with

institutional policies and underlying resources.

VIDaaS Admin Node typically resides on its own

virtual node. In a production environment, it may be

desirable to have several VIDaaS Admin Nodes to

provide redundancy and resilience. The VIDaaS

admin node itself consists of three sub-components

i.e. VIDaaS Admin Portal; VIDaaS Admin Database;

and RESTFul Interface.

2.1.1 VIDaaS Admin Portal

The VIDaaS Admin Portal is the main Web interface

to access the whole system either as VIDaaS users or

VIDaaS administrator. The VIDaaS Admin Portal

allows VIDaaS users to administrate projects and

databases owned by them and it also provides an

interface for system administrators to configure and

manage the whole VIDaaS system.

VIDaaS registered users can create new projects,

administrate all projects owned by them, and view

projects for which they have access rights. Once the

user selects any specific project he/she is redirected

to the project’s own virtual machine/node (section:

2.4). Tasks related to any specific project i.e.

creating, modifying, deleting and mirroring the

database are performed on the project’s node.

2.1.2 VIDaaS Admin Database

The VIDaaS Portal stores details of registered users

and successfully created project in the local database

hosted on the VIDaaS Admin node. The admin

database stores all information about registered

users; deployed projects, project members and their

roles; and databases within each project. The

VIDaaS Admin Database links users with their

organisational ‘Single Sign On’ facilities; maps users

with projects and databases; automates projects

backup and manages project/database recovery from

system level failure.

The single place of storage for system

information can be potentially single place of failure

and can also result in a bottleneck. On the contrary,

it also improves easy system management,

monitoring, administration and the report generation.

The future version of VIDaaS will support

multiple Admin Nodes, each with its own local

database for load balancing and resilience.

VIDaaS will also provide out of box support to

synchronise all VIDaaS Admin Nodes and their

local databases.

2.1.3 RESTful Interface

The RESTful Interface of the VIDaaS admin node is

the heart of VIDaaS level inter-component

communication within the system. The VIDaaS

Admin Node interacts with Identity Access

Management either when user registers with VIDaaS

or sign-in. Similarly through RESTful Interface, the

VIDaaS admin node communicates with

Virtualization Node and Billing and Payment Node

whenever a new project is created, existing projects

is modified or existing project is dismantled. The

communication between VIDaaS Admin Node and

Project Node/s takes place whenever the project

database/s are created, modified or deleted. Finally,

VIDaaS Admin Node communicates with the

Backup Node at regular intervals to trigger the

VIDaaS-DesignArchitectureofVirtualInfrastructurewithDatabaseasaService

103

Figure 2: Communication Sequence Diagram between various components.

project’s database backup. The simplistic use case of

a user registration, login, project creation, project

database creation and project database backup in the

form of sequence diagram in shown in the Figure: 2.

This top level sequence diagram highlights

interaction between various VIDaaS components

according to their roles.

2.2 Identity Access Management Node

The Identity and Access Management Node is the

core of protecting user data from un-authorised

access. The VIDaaS system authenticates users

through Identity Access Management node so that

only authorised users are able to access the services.

The VIDaaS system facilitates database sharing and

collaboration and supports various user roles within

the system

User roles determine how the user authenticates

with VIDaaS and what features of the VIDaaS

system are accessible to the user.

It is the job of the IAM component of the

VIDaaS project to enable a user to access VIDaaS

services in as simple and unobtrusive way as

possible whilst, at the same time, keeping all their

project data completely secure from unauthorised

access. The IAM implementation within VIDaaS

needs to allow secure access to the VIDaaS system

with the minimum interaction from administrators

thus automating most of the service or initiated by

the end user.

2.3 Virtualization Node

The Virtualization Node encapsulates features like

creating, cloning and dismantling individual virtual

machines. The Virtualization Node has been

separated from the VIDaaS Admin node to aid with

the abstraction of VIDaaS from the underlying

virtualisation technologies. In current release the

Virtualization Node only supports VMware vSphere

Hypervisor (based on ESXi) with possibility of other

hypervisor technologies such as Xen, KVM, Hyper-

V, etc. The Virtualization Node has following three

components:

 VMWare Adapter: The Virtualization Node

accepts requests from the VIDaaS Admin

Node to create and tear down projects. Since

each project resides in its own virtual

machine, when a new project is created by a

user, it is the job of the Virtualization Node to

create a new virtual machine and seed it with

the relevant software to enable user access via

VMWare adapter.

 REST Interface: In VIDaaS the Virtualization

Node is only accessed via VIDaaS Admin

Node to minimize potential security risks. All

communication between VIDaaS Admin Node

and the Virtualization Node is through REST

Interface which itself uses secure sockets with

RESTful Web services.

 VMWare Management Portal: Virtualization

Node also has Web interface to interact with

underlying hypervisor technology to create,

modify and dismantle individual virtual

machines. The Web interface to manage

virtual machines is only available to VIDaaS

administrators, though in future release all this

functionality will only be available from the

VIDaaS Admin Node.

CLOSER2012-2ndInternationalConferenceonCloudComputingandServicesScience

104

The separation of Virtualization node from other

components within VIDaaS deployment architecture

can lead to potential support of various hypervisors

in parallel to each other. Though the Virtualization

Node is currently using VMware vSphere

Hypervisor but the underlying machines are created

in Open Virtualization Format (OVF). The OVF

standard is not tied to any particular hypervisor or

processor architecture.

2.4 Project Node

Each project resides in its own virtual machine

called the Project Node. The Project Node is

automatically created and made publically available,

once the user creates new project via the VIDaaS

Admin Node. Users can own more than one project

and each project will reside in its own virtual

machine. The Project Node has following three sub-

components:

 Web interface;

 Database; and

 RESTful Web service.

The Web interface itself is logically split in two

interfaces:

 Project Admin Interface: to create database/s;

backup database/s; or delete database/s. The

Project Admin interface also allows to add or

remove other users (already registered with

VIDaaS) as project members, update their role

within the project; and

 Project Interface: to manage data in each

individual database i.e. access, query and

update the data within the database. The

Project interface is only available when

project owner’s want their database/s

publically available. The Project Interface

reverse engineers the database and creates

appropriate set of web pages for each table

within the database. The generated web pages

have a form to add new rows in the table,

query database and display table contents in

tabular format.

The virtual machine for each project has an

instance of a PostgreSQL, installed by default. Users

create

empty database in the PostgreSQL installation

via Project Admin interface.

The VIDaaS project is designed to cater two

potential use cases, i.e. existing research groups with

working databases and new research groups without

databases. Research groups with working database

can populate database in various ways either by

uploading the existing Microsoft Access database,

database in comma separated value format or the

database dump from the existing database. The

VIDaaS itself supports various versions of Microsoft

Access database in a fairly uniform way.

The VIDaaS project has also developed Web

based SQL Designer to design the database from

scratch. The SQL Designer is developed using

various JavaScript libraries and fully support drag

and drop. The purpose of SQL Designer is not only

to design the database from scratch but also modify

the existing databases held by VIDaaS. Though,

users can use their preferred existing desktop tools

to modify the database and upload the modified

database, but we encourage the use of SQL

Designer.

The VIDaaS project also has an experimental

tool not yet fully integrated to create the project

database from publically available database. This

tool pulls the database structure and its data from

publically available databases and mirrors the

gathered information as a local database. This

experimental tool has support for different popular

databases. This tool can also be used to mirror

databases hosted within the VIDaaS to external

servers.

The VIDaaS project specifically aims towards

research groups and is designed with the notion

changes in the research database will be required.

VIDaaS introduces the concept of three versions of a

single database i.e. live, old and test database. The

newly created database by user is always a live

database of a project within the VIDaaS system.

When researchers plan significant modification to

the original database (changes in the structure of the

original database), then they have the option to

create the “test” database by a single click. The

“test” database is pure mirror of “live” database. The

planned changes may make the database

incompatible with existing database due to addition

or removal of tables or columns, changes in tables

and columns names, changes in columns data types

or even due to changes in relations between various

tables. These potentially incompatible changes

highlight the significance of the “test” version of the

database. Once researchers are happy with changes

in the “test” database, they can easily make the

“test” database into “live” database. Similarly, on

reaching any major milestone researchers are

suggested to make a backup copy of their live

database as an “old” database. The manual mirroring

of the database as “old” database is independent of

the VIDaaS system level regular backups done

according to the project configurations. Similarly, if

the project’s “live” database has data loss, data

VIDaaS-DesignArchitectureofVirtualInfrastructurewithDatabaseasaService

105

corruption or any other unexpected problem, project

owners can recover the database from the previous

milestone saved as an “old” database. The figure 3

shows possible transitions between different

versions of a single database.

Figure 3: Three potential versions of a database.

The RESTful interface of the Project node is

mainly to communicate with VIDaaS Admin node.

The communication is to update the VIDaaS Admin

Node of changes made in the project. These changes

include changes in project configurations, members

and project databases i.e. creating or deleting

database, creating or deleting different versions of

database. Any change related to database design or

data contained within the database is not propagated

to VIDaaS Admin Node. The RESTful interface is

also used to initiate the database recovery or backup

from/to the Backup Node (discussed in next

section).

2.5 Backup Node

The Backup Node is a system level backup facility

of all databases managed by VIDaaS. The Backup

Node is designed with elasticity in the mind. The

VIDaaS architecture has provision to manage

multiple Backup Nodes as demand grows. In

VIDaaS, Backup Nodes are added to the system as

required and are dismantled automatically when they

are not required. Though, at any given time there is

always one Backup Node within the VIDaaS.

The purpose of Backup Node is to provide

recovery mechanism from the VIDaaS system level

failures. The Backup Node also provisions for

human mistakes i.e. project owners/administrators

mistakenly deleted database, or wiped out all tables

within the database. The system level backup keeps

the backup of “live” version of each database hosted

by VIDaaS on the Backup Node. The VIDaaS

backup is triggered automatically based on the

project configurations and Service Level

Agreements (SLA).

As described earlier, each project is created by

accessing VIDaaS Admin Portal and only VIDaaS

Admin node has access to each registered project.

The project owners select the backup policy of the

database at the project registration time. Currently,

VIDaaS only supports handful backup policies i.e.

daily, weekly, fortnightly or monthly. The VIDaaS

Admin node instantiates the system level backup of

the project database according to the project

configurations. The Backup Node doesn’t run any

live database and only archives the database dump in

the flat file format.

2.6 Pricing and Payment Node

The Pricing and Payment Node, as name suggests is

the charging module of VIDaaS. The Pricing and

Payment Node is responsible for charging projects

according to their usage of VIDaaS resources and

processing the payment. This module is not yet

developed and deployed.

3 VIDaaS TOOLS

VIDaaS has developed few tools to complement

Database as a Service. These tools provide one stop

solution to all database needs of researchers in the

form of VIDaaS. Tools developed for VIDaaS are

standard based and can be used either as a

standalone application or as a VIDaaS integrated

component. Few of the main tools integrated with

VIDaaS are explained below:

3.1 SQL Designer

VIDaaS provides the tool to design the database

from scratch to researchers starting new projects.

The SQL Designer is user friendly, easy to use,

innovative; standards compliant designer with drag

and drop support. The SQL Designer can be

effectively used by researchers with basic

understanding of Structured Query Language (SQL).

The SQL Designer also supports different type of

relations between tables i.e. one-to-one, one-to-

many, many-to-many, imposing and non-imposing.

Columns within the table are represented by

different colours to indicate primary, foreign and

ordinary keys. The current version of SQL Designer

produces SQL compliant with PostgreSQL. Tthe

future will capable to generate SQL compliant with

various popular database systems i.e. Oracle,

MySQL, SQL Server, IBM DB2 etc. and users will

have a choice to select the database of their choice.

3.2 Database Migration Tool

The VIDaaS also provides tool to migrate existing

databases into the VIDaaS with minimum hassle.

The database migration tools support most of

popular databases available in the market. The only

CLOSER2012-2ndInternationalConferenceonCloudComputingandServicesScience

106

requirement of this tool is the publically accessible

database.

3.3 Form Builder

The VIDaaS also provides the tool to create the

HTML form with drag and support. The purpose of

the Form Builder is to create a form to support

complex database queries from multiple tables.

Project administrators are responsible for creating

such forms and making them available along with

project portal. Users will fill the form created

through the Form Builder and users entered values

will be translated into SQL queries.

3.4 Supporting Microsoft Access

The support for Microsoft Access database was

crucial for the success VIDaaS and its wider

acceptance within the research community of Arts

and Humanity. VIDaaS has robust and easy to use

support for various versions of Microsoft Access

database. Researchers are only required to upload

their existing databases to the VIDaaS. The

uploaded Microsoft Access database is mirrored as a

PostgreSQL database with a single click. The

mirroring process maintains relational and data

integrity of the uploaded Microsoft Access database.

4 FUTURE WORK

VIDaaS system is based on single tenancy database;

which is simpler to implement and manage but is

burden on cloud resources. For, future versions will

investigate multi-tenancy database arrangement;

where similar research databases share the same

virtual machine. The multi-tenancy arrangement

leads to the concept of “Community Cloud”. The

community cloud is used and controlled by a group

of users with similar and shared interests. In our case

community cloud means researchers from same

domain tackling similar problem.

The Billing and Payment Node of VIDaaS is in

the design stage and needs to be developed, tested

and integrated with rest of the system.

The support for Document Based databases such

as CouchDB or MangoDB is missing from VIDaaS

and needs to be added. The aim of VIDaaS is an

umbrella project to commission different types of

databases on demand in cloud environment. We also

need to bring diversity in our test users rather than

confining VIDaaS to humanity and social science

domain.

5 CONCLUSIONS

This paper outlines the architectural design and

deployment strategy adopted for database as a

service, in the VIDaaS project. The VIDaaS

architecture is modular in nature, with different

components encapsulating core functionalities in

accordance with the Service Oriented Architecture.

The modular design of VIDaaS has benefits during

the development, deployment and testing stage of

the whole project. The paper also briefly discusses

various tools developed to enhance the effectiveness

of the VIDaaS framework.

REFERENCES

Amazon.com, Inc. 2011, Amazon SimpleDB, viewed 12

December 2011, https://aws.amazon.com/simpledb/

Supporting Data Management Infrastructure for the

Humanities, viewed 12 December 2011, http://

sudamih.oucs.ox.ac.uk/

Amazon.com, Inc. 2011, Amazon Relational Database

Service (Amazon RDS), viewed 12 December 2011,

https://aws.amazon.com/rds/

Virtual Infrastructure with Database as a Service, viewed

12 December 2011, http://vidaas.oucs.ox.ac.uk/

Chang F. 2006, 'Bigtable: A Distributed Storage System

for Structured Data', OSDI'06: Seventh Symposium on

Operating System Design and Implementation,

http://research.google.com/archive/bigtable.html

DabbleDB (now purchased by twitter), Website not

available on 12 December 2011

Internet2 2011, Shibboleth project, viewed 12 December

2011, http://shibboleth.net/

Logicalis Group 2011, viewed 12 December 2011,

http://www.logicalis.com/

Martinez-Uribe, L 2009, Supporting Data Management

Infrastructure for the Humanities (Sudamih), Oxford

University Computing Services, Oxford, viewed 12

December 2011, http://sudamih.oucs.ox.ac.uk/

Microsoft Corporation 2011, SQL Azure, viewed 12

December 2011, http://www.windowsazure.com/

en-us/home/tour/database/

Salesforce.com 2011, database.com, viewed 12 December

2011, http://www.database.com/

Ragouzis, N 2008, Security Assertion Markup Language

(SAML) V2.0 Technical Overview. OASIS Commit

tee Draft, Document ID sstc-saml-tech-overview-2.0-

cd-02 http://www.oasis-open.org/committees/

download.php/27819/sstc-saml-tech-overview-2.0-cd-

02.pdf

VMware, Inc. 2011, Palo Alto, viewed 12 December

2011, http://www.vmware.com/

VIDaaS-DesignArchitectureofVirtualInfrastructurewithDatabaseasaService

107