Migrating to the Cloud - A Software Product Line based Analysis

Abstract

Abstract:Identifying which part of a local system should be migrated to a public Cloud environment is often a difficult and error prone process. With the significant (and increasing) number of commercial Cloud providers, choosing a provider whose capability best meets requirements is also often difficult. Most Cloud service providers offer large amounts of configurable resources, which can be combined in a number of different ways. In the case of small and medium companies, finding a suitable configuration with the minimum cost is often an essential requirement to migrate, or even to initiate the decision process for migration. We interpret this need as a problem associated with variability management and analysis. Variability techniques and models deal with large configuration spaces, and have been proposed previously to support configuration processes in industrial cases. Furthermore, this is a mature field which has a large catalog of analysis operations to extract valuable information in an automated way. Some of these operations can be used and tailored for Cloud environments. We focus in this work on Amazon Cloud services, primarily due to the large number of possible configurations available by this service provider and its popularity. Our approach can also be adapted to other providers offering similar capabilities.

Constraint programming heuristics for configuring optimal products in multi product lines

“…Current CP-based approaches support the configuration of products for a PL by considering single-objective optimization [8,9,20,24,41] , and multi-objective optimization [16,29,35] . These approaches have evaluated performance and/or scalability concerns, by considering the number of configuration constraints and the size of the related FM.…”

“…The execution time to find a configuration varies depending on the considered constraints and the size of the PL. For all cases, solutions are obtained in at least 1.8 s [20] and at most 1 h 30 min [9] . However, if the complexity and size of the FM increase, or the number of considered configuration constraints is higher, these values can be drastically affected.…”

“…However, if the complexity and size of the FM increase, or the number of considered configuration constraints is higher, these values can be drastically affected. In fact, for the FM based on the AWS EC2 service, which has a total of 35 features and 5 configuration constraints, the execution time was around 1 h 30 min for generating 4 configurations [9] .…”

“…This is especially important when the introduced non-functional requirements are translated into optimization problems where a predefined set of numerical non-functional values related to product features should be minimized or maximized. To face this challenge, some approaches based on constraint programming have been developed to solve single-objective optimization [8,9,20,24,41] , and multi-objective optimization [16,29,35] problems. These approaches provide a product line configuration in between 1.8 s [20] and 90 min [9] depending on the product line size (models with up to 50 0 0 features), the number of optimization objectives (e.g., 1 to 5), and other confounding variables that are meant to be explored and discovered by researchers in the field (e.g., type of involved constraints).…”

Evolutionary composition of QoS-aware web services: A many-objective perspective

“…the piece of a service contract where the level of service is determined, could even bring more alternatives into play, considering that a specific service provider might offer several QoS configurations for the same service provision. For example, Amazon Web Services (AWS) establishes up to 8 different deployment plans (instances) 15 for their computing services (EC2) (Wada et al, 2012), which combined with other configurable options like the operating system, the available CPU and backup settings can lead to 16,991 possible configurations (García-Galán et al, 2013). Although highly demanded because of its flexibility, considering a larger set of configuration alternatives implies increasing the number of QoS properties 20 coming into play and, consequently, finding appropriate trade-offs among them becomes extremely difficult.…”