13.1 Introduction
13.2 Context and Motivation
13.3 Application Scenario
13.3.1 Cloud Provider Selection
-
The cost of re-evaluating the providers on a regular basis in order to improve quality of service and cost efficiency would be very high.
-
The goal of covering different markets will require addressing strict data location policies and therefore the ability to deploy to multiple Clouds. These aspects are hard to address using the initial approach since it would require a much larger number of providers to be analysed resulting in very high efforts in data gathering and analysis.
-
The criteria chosen intuitively based on BOC’s own experience were not comprehensive enough to cover all relevant aspects.
-
In particular, the ease of migration from one provider to another, which should have been considered as one of the most important criteria, was ignored.
Stakeholders | Type of intangible asset | Assets considered |
---|---|---|
Business representatives | Business oriented intangible assets | Customer loyalty, legislation compliance, internal efficiency and performance, sales rate, market awareness, improve product innovation and quality |
Technical (DevOps) representatives | Technology oriented intangible assets | Data privacy, data integrity, maintainability, end user performance, service availability, cost stability |
13.3.2 Application Deployment to Multiple Clouds
-
Even though Puppet provides configuration modules for different Cloud stacks like e.g. OpenStack, provisioning of IaaS instances in a totally Cloud-stack-independent, transparent way was hard to accomplish.
-
Deploying parts of the application (e.g. the database tier) to PaaS would be even harder since it would require the use of another configuration to deploy the application stack to PaaS.
-
It would enable them to automatically deploy to any of the Cloud systems previously selected by Venues 4Clouds including the provisioning of Cloud instances in a transparent way, resulting in a higher degree of automation and consequently less manual operation efforts.
-
The model-based nature of the approach was expected to enable BOC to document individual deployments in a traceable and comprehensible way and support them in explaining deployment decisions to their customers.
13.3.3 Cloud Application Monitoring
13.3.4 Cloud to Cloud Migration
13.4 Conclusion and General Recommendations
-
When selecting a particular Cloud service, the ease of migration to another equivalent service should be considered. This implies on one hand the existence of such services and on the other hand the ability to migrate software components and data to these other services easily.
-
In order to increase cost efficiency and quality of service the Cloud service provider market should be analysed on a regular basis. The selection of Cloud services and Cloud providers might become a reoccurring task. A common knowledge base and a tool based approach for decision making as planned for MODAClouds Venues 4Clouds tool will help saving efforts for data acquisition and analysis and making decisions in a traceable, comprehensible way.
-
In order to be able to easily deploy to different providers deployment automation or even self-adaptation should be considered to save operation efforts. Automated deployment should ideally work on different Cloud stacks (i.e. on different Cloud services) with as little adaptations as possible.
-
Monitoring should be considered an integral part of the Cloud service as it is the only reliable way to track SLA adherence. The solution should be easy to use, maintain, extend and at best it shall be managed together with the business application by means of the configuration management system. The combination of an established product such as Icinga with the sophisticated Tower 4Clouds RDF stream processing toolkit is a good candidate for this challenge.