A Survey on Database Performance in Virtualized Cloud Environments

A Survey on Database Performance in Virtualized Cloud Environments

Todor Ivanov (Technische Universität Darmstadt, Germany), Ilia Petrov (Technische Universität Darmstadt, Germany) and Alejandro Buchmann (Technische Universität Darmstadt, Germany)
Copyright: © 2012 |Pages: 26
DOI: 10.4018/jdwm.2012070101
OnDemand PDF Download:
$37.50

Abstract

Cloud Computing emerged as a major paradigm over the years. Major challenges it poses to computer science are related to latency, scale, and reliability issues. It leverages strong economical aspects and provides sound answers to questions like energy consumption, high availability, elasticity, or efficient computing resource utilization. Many Cloud Computing platform and solution providers resort to virtualization as key underlying technology. Properties like isolation, multi-virtual machine parallelism, load balancing, efficient resource utilization, and dynamic pre-allocation besides economic factors make it attractive. It not only legitimates the spread of several types of data stores supporting a variety of data modes, but also inherently requires different types of load: (i) analytical; (ii) Transactional/Update-intensive; and (iii) mixed real-time feed processing. The authors survey how database systems can best leverage virtualization properties in cloud scenarios. The authors show that read mostly database systems and especially column stores profit from virtualization in analytical and search scenarios. Secondly, cloud analytics virtualized database systems are efficient in transactional scenarios such as Cloud CRM virtualized database systems lag. The authors also explore how the nature of mixed cloud loads can be best reflected by virtualization properties like load balancing, migration, and high availability.
Article Preview

Introduction

Cloud Computing is a major paradigm that emerged over the last half a decade and has already impacted industrial products and services; besides Cloud Computing represents an active research topic. It is gaining such a significant momentum because of a mixture of economical and technical factors. In economical terms, flexible billing, reduction in TCO and administration costs, license costs and so on count to the advantages that Cloud Computing offers. In 2009, the National Institute of Standards and Technology (NIST) (Mell & Grance, 2009) presented a definition of Cloud Computing:

Cloud Computing is a model for enabling convenient, on-demand network access to a shared pool of configurable computing resources (e.g., networks, servers, storage, applications and services) that can be rapidly provisioned and released with minimal management effort or service provider interaction.”

Exactly this provisioning of various different resources and services makes it difficult to formally define what Cloud Computing is. Currently, there is even a growing trend to offer each layer of the enterprise architecture stack, starting from the hardware and up to the application layer, as a separate service also called *aaS or XaaS(Everything as a Service) (Bose, 2008). The three main types of cloud services that are becoming a standard are: Software as a Service (SaaS), Platform as a Service (PaaS) and Infrastructure as a Service (IaaS). IaaS offers direct access to server infrastructure based on the customer resource demands. The physical location of the system is transparent to the customer and the cloud provider takes care of the hardware maintenance. One of the biggest infrastructure cloud providers are Amazon (S3, SimpleDB and EC2), Rackspace (http://gogrid.com) (Cloud Hosting and Cloud Storage) and IBM (Smart Business Storage Cloud and Computing on Demand). PaaS is the next level of service, where a software platform with development tools and frameworks is offered. This environment allows the customer to easily develop, run and host their own online applications like SaaS. Popular cloud platforms are Google AppEngine, Microsoft Azure and Salesforce platform. SaaS offers the customer direct access to applications through the internet without the need to preinstall them on a computer or any smart device, which makes the service device-independent. Examples for such applications are online office suites (Google Docs, IBM LotusLive iNotes, Zoho and Microsoft office suites), Google calendar, photo galleries (Google Picasa, Yahoo Flickr, etc.) and many others. All these services enable the customers to pay and use only the services and resources that they need, which makes Cloud Computing dynamically scalable. It saves them money and time, as they do not have to build and maintain their own platform infrastructure. On the other hand, Cloud Computing is task and data centric as the users can focus on their core business, by extending the functionality of their product and even use additional cloud tools to speed up their productivity. Virtualization has become an essential part of every data center infrastructure and basic building block of Cloud Computing, automating task and data management.

In this paper we survey how database systems can best leverage virtualization properties in cloud scenarios. Many authors show that virtualized database systems need special tuning to handle cloud workloads efficiently. We show that read-mostly database systems and especially column stores profit from virtualization in analytical and search scenarios. Secondly, for cloud analytics virtualized database systems are efficient as well. Last but not least, in transactional scenarios such as Cloud CRM virtualized database systems lag by 10%-15% which is acceptable given the administrative benefits. We also explore how the nature of mixed cloud loads can be best reflected by virtualization properties such as load balancing, migration and high availability.

Complete Article List

Search this Journal:
Reset
Open Access Articles: Forthcoming
Volume 13: 4 Issues (2017)
Volume 12: 4 Issues (2016)
Volume 11: 4 Issues (2015)
Volume 10: 4 Issues (2014)
Volume 9: 4 Issues (2013)
Volume 8: 4 Issues (2012)
Volume 7: 4 Issues (2011)
Volume 6: 4 Issues (2010)
Volume 5: 4 Issues (2009)
Volume 4: 4 Issues (2008)
Volume 3: 4 Issues (2007)
Volume 2: 4 Issues (2006)
Volume 1: 4 Issues (2005)
View Complete Journal Contents Listing