IntroductionBackground and Related WorkService Deployment ProcessDeployment Process OverviewDeployment Model and Problem DefinitionThe Cyberaide Project and Cyberaide CreativeThe Cyberaide ProjectCyberaide Creative ParadigmEmergency Services DirectoryOverviewDeployment Strategy OverviewDeployment SolutionSolution EvaluationFuture Work and ConclusionReferencesTowards On Demand IT Service DeploymentJai Dayal2, Casey Rathbone2, Lizhe Wang1, and Gregor von Laszewski11Pervasive Technology Institute, Indiana University2729 E 10th St., Bloomington, IN 47408, U.S.A.2Service Oriented Cyberinfrastruture Lab, Rochester Institute of TechnologyBldg 74, Lomb Memorial Drive, Rochester, NY 14623-5608Email corresponding author: [email protected]—Advanced IT solutions allow users to create, orga-nize and share user services and computing resources across awide range of heterogeneous platforms. This makes managing(updating and deploying) the IT systems considerably morecomplex. Management tasks are so entangled within the existingcomponents that it requires a IT professional with advancedtraining and intimate knowledge of the existing configuration tocomplete them.In this chapter, we will present the deployment solution for alive complex IT system, the Emergency Services Directory (ESD).Many different technologies exist that attempt automate theapplication deployment process by allowing a user to describe thedependencies, provide the configuration parameters, and specifythe application’s required technologies, such as the operatingsystem, database or Web server technology.ESD’s deployment solution takes advantage of the benefitsprovided by virtualization, and virtual appliances in particular.Each of ESD’s components are wrapped and contained with in avirtual appliance image. To automatically distribute and deploythe virtual appliance image on-demand, we utilize the CyberaideCreative tool, which is a tool being developed in the on-goingCyberaide project.I. INTRODUCTIONTypically, an IT service or application consists of severalcomponents inter-operating to perform a set of functionsor tasks. Service deployment is considered the process ofacquisition and execution of the service, i.e., making theservice ready for use [1]. Deployment typically consists ofthe release of the software, the configuration of the software,and the installation of the software [2].Depending on the application, the deployment process canbe quite complex requiring a user or developer to have adetailed understanding of the applications individual compo-nents. An IDC survey states that out of the $95 billion dollarsspent on application operating costs, 19% can be attributed tothe cost of deployment alone [3].With today’s advanced IT infrastructures, such as computeGrids and Clouds, many users, who would benefit from theirusage, avoid these technologies due to their high level ofcomplexity and steep learning curve. The complexity andsteep learning curve of these systems is largely due to thevast heterogeneity of the components and lack of more user-friendly software tools. Applications require specific operatingsystem drivers and configurations as well as software librariesand packages for deployment [1]. In heterogeneous environ-ments, access to resources matching the application’s specificrequirements can not always be guaranteed. Therefore a usermay be required to provide several version of the applicationto match the infrastructure’s different resources which couldbe too cumbersome a task within a large environment [4].To help assuage the problems within heterogeneous environ-ments, users turn to virtual machine (VM) technology. VMsoffer users [5]:• Customized OS: The user has great flexibility to cus-tomize the operating system to meet the application’srun-time requirements. These customizations can include,but are certainly not limited kernel level customizationsto memory and storage customizations.• Ease of Management: Virtual Machines can easily beshutdown or restarted, easing the system reconfigurationprocess. Additionally, VMs can easily be migrated todifferent physical machines, thus allowing the applicationto easily operate during a physical machine’s downtime.• System Security: VM users can define the operatingsystem’s privilege with out affecting the privileges of theunderlying operating system. For example, root access isoften required to install operating system modules. Instal-lation of these modules only effect the VM. Additionally,if a configuration causes the system to crash, only the VMis affected, while the underlying machine and operatingsystem remain operational.While VMs provide us with a flexible and easy to deployplatform, VMs alone fall short of automatic and on-demandservice deployment. For example, each time the VM is shut-down, the service will have to be re-deployed.Virtual appliances take VMs a step further by containingboth the platform and the service. Launching and executingvirtual appliances, however, still requires the user to un-derstand many technical details. For example, a user mustunderstand which basic software packages or components areneeded for the application. The user must also understand thefunctional dependencies between these packages. For example[1] stats that IBM DB2 has approximately 40 configuration pa-rameters that must be resolved during the deployment process.Our contribution uses the Cyberaide Creative tool to allow auser to select the appropriate virtual appliance image, and han-dles the resource selection, resolves the dependencies betweencomponents and packages, and automates the deployment,shielding the user from the process. Using ESD as a casestudy, we evaluate Cyberaide Creative’s success.The rest of this chapter is organized as follows. SectionII presents some background information and compares our2solution to the existing solutions in the field. Section IIIdiscusses and formulates the deployment process. Cyberaideand Cyberaide Creative are presented in Section IV. SectionV describes the ESD project, our case study, and Section VIpresents our deployment solution. Section VII evaluates oursolution and our conclusions are presented in Section VIII.II. BACKGROUND AND RELATED WORKAutomatic and on-demand service deployment is not anew idea there currently exist several tools to facilitate thedeployment process. A common one is the package managersfound in various Linux distributions such as Fedora andUbuntu. These tools consist