PaperlayerView exact revision on arXiv

Source-linked AI summary

CloudGenius: Decision Support for Web Server Cloud Migration

Michael Menzel, Rajiv Ranjan

arXiv:1203.3997v1cs.DCcs.SE

TL;DR

Migrating Web applications to the Cloud requires selecting compatible VM images and infrastructure services across heterogeneous criteria and dependencies. CloudGenius automates this decision process with a model, user preferences, and multi-criteria evaluation, and its prototype was validated while the algorithm’s time complexity was analyzed.

  • Problem

    Web-cloud migration requires manually difficult selection of compatible VM images and infrastructure services across heterogeneous criteria and complex dependencies.

  • Method

    CloudGenius uses a model, user requirements and preferences, and Analytic Hierarchy Process-based multi-criteria decision-making to automate selection.

  • Results

    The prototype was validated through experiments, and the underlying algorithm’s time complexity was analyzed.

  • Takeaways & Limitations

    CloudGenius provides an applicable decision-support approach for selecting VM images and infrastructure services within a Web-application migration process.

  • Takeaways & Limitations

    CloudGenius focuses on single-tier Web applications, restricts criteria to quantitative numerical factors, and assumes full-featured VM images.

Abstract

from arXiv · show

Cloud computing is the latest computing paradigm that delivers hardware and software resources as virtualized services in which users are free from the burden of worrying about the low-level system administration details. Migrating Web applications to Cloud services and integrating Cloud services into existing computing infrastructures is non-trivial. It leads to new challenges that often require innovation of paradigms and practices at all levels: technical, cultural, legal, regulatory, and social. The key problem in mapping Web applications to virtualized Cloud services is selecting the best and compatible mix of software images (e.g., Web server image) and infrastructure services to ensure that Quality of Service (QoS) targets of an application are achieved. The fact that, when selecting Cloud services, engineers must consider heterogeneous sets of criteria and complex dependencies between infrastructure services and software images, which are impossible to resolve manually, is a critical issue. To overcome these challenges, we present a framework (called CloudGenius) which automates the decision-making process based on a model and factors specifically for Web server migration to the Cloud. CloudGenius leverages a well known multi-criteria decision making technique, called Analytic Hierarchy Process, to automate the selection process based on a model, factors, and QoS parameters related to an application. An example application demonstrates the applicability of the theoretical CloudGenius approach. Moreover, we present an implementation of CloudGenius that has been validated through experiments.

1. INTRODUCTION

CloudGenius addresses the complexity of migrating Web applications by supporting compatible infrastructure-service and VM-image selection. It organizes migration into a process intended to reduce engineers’ decision effort.

  • Cloud migration decisions must weigh factors including price, SLA level, support quality, security, costs, risks, and service-level expectations.
  • VM-image selection is complex because images differ in operating systems, software stacks, versions, and compatibility with infrastructure services.
  • The migration process includes infrastructure-service selection, VM-image selection and customization, migration-strategy definition, and strategy application.
  • The process order reflects dependencies between images and infrastructure services, while multi-component migrations additionally require considering interconnections.
  • CloudGenius provides a detailed process and decision support intended to reduce the effort of finding suitable infrastructure services and VM images.

2. RELATED WORK

Prior work addresses parts of Cloud service selection and migration, but does not jointly provide Web-server and VM-image selection with decision support and a migration process.

  • Ontology-based approaches can select VM images and infrastructure services but lack service evaluation, preventing quality and suitability differentiation among qualifying services.
  • The Cloud Adoption Toolkit supports whole-enterprise IT-infrastructure migration through risk management and workload-based cost calculation.
  • Existing Cloud service-selection work lacks integration of actual Web-server and VM-image selection with decision support and a migration process.
  • Other approaches evaluate or measure Cloud-provider services, but provider-selection methods are missing in the cited performance-capability work.

3. CLOUDGENIUS FRAMEWORK

CloudGenius formulates compatible VM-image and infrastructure-service selection as multi-criteria decision problems, using user requirements and preferences to identify a combined solution. Its model represents alternatives, attributes, requirements, criteria, and compatibility dependencies, while its process supports iterative migration decisions.

  • CloudGenius Framework: CloudGenius translates VM-image and infrastructure-service selection into multi-criteria decision-making problems and joins them into a combined feasible solution.
  • CloudGenius Framework: The framework uses the (MC2)2 evaluation process and Analytic Hierarchy Process to structure requirements, criteria, alternatives, and weighted evaluations.
  • CloudGenius Framework: Users define an abstract Web server, acceptable requirements, relevant criteria, and pair-wise preference weights that guide image and service recommendations.
  • Migration Process: The migration process can loop through worse alternatives or altered requirements until a satisfying solution is found and the application is migrated.
  • Formal Model: The formal model contains images, services, providers, attributes, requirements, goals, and leaf criteria for evaluating alternatives.
  • Formal Model: CloudGenius selects the highest-valued feasible image-service combination, with compatibility dependencies explicitly represented because providers may not support standard formats.
  • Formal Model: The approach depends directly on accurate, current model data, requiring measurements and database maintenance to support result quality.

3.3 Web Server Requirements & Preferences

CloudGenius lets engineers specify Web server requirements and weighted preferences, then evaluates compatible VM images using AHP-based criteria and requirement checks. The process ranks eligible images by their aggregate value while supporting extensible goal hierarchies and attributes.

  • Requirements: Engineers define numerical minimum or maximum requirements and allowed non-numerical values for the target Web server.Requirement types include Max, Min, Equals, and OneOf conditions.
  • Preferences: Preferences are expressed by weighting goals and criteria through pair-wise comparisons modeled on AHP.CloudGenius provides goal hierarchies for VM images and infrastructure services.
  • Preferences: The VM image hierarchy prioritizes cheapest and best-quality images using hourly price and image popularity.The proposed hierarchy remains extensible to additional QoS goals and more complex structures.
  • Image evaluation: VM images are evaluated as alternatives using numerical and non-numerical attributes, with compatibility and virtualization constraints affecting eligibility.A VM image must be deployable on the selected infrastructure service, and images failing requirements are eliminated from composite solutions.
  • Scope and maintenance: The image attribute catalog is extensible, but result quality depends directly on the currentness and quality of stored attribute data.The paper identifies attribute selection and maintenance as ongoing concerns.
  • Image evaluation: The AHP-derived evaluation function assigns each image a value, returning zero when its attributes fail all specified requirements.Images are ranked by evaluation value, with fallback evaluation when no image satisfies every requirement.

3.5 Cloud Infrastructure Service Selection

CloudGenius evaluates infrastructure services in parallel with VM images using numerical and non-numerical service attributes within an AHP-based decision method. Service alternatives are ranked after requirements, criteria weights, and attribute data are incorporated into the evaluation.

  • Evaluation method: Infrastructure service selection uses numerical and non-numerical attributes evaluated through the (MC2)2 framework.The resulting service evaluation function is g(s_j, Â_sj, B̂_sj) → v_sj.
  • Service attributes: Service attributes include dynamic numerical measures with defined metrics and value ranges, alongside static non-numerical example values.The numerical and non-numerical attribute sets are summarized in Tables 5 and 6.
  • Service attributes: Hourly usage prices may require calculation from provider pricing models and correction into comparable total- or monthly-cost metrics.Cost comparability can require renaming or adjusting the price attribute.
  • Ranking: After alternatives, hierarchy, requirements, and weights are set, AHP creates the service evaluation method and produces values used for ranking.The highest-ranked service is selected from the evaluated alternatives.

3.6 Best Combination

CloudGenius builds feasible image-service combinations, evaluates them with a compensatory function, and selects the highest-valued solution. An integrated approach is possible, but the framework recommends separate selection because the trade-off is more transparent.

  • Combination construction: CloudGenius filters infeasible image-service pairs using dependency set D before evaluating combined solutions.Each image-service pair represents a possible solution, but only compatible compositions remain feasible.
  • Combination evaluation: The combination value combines image and service evaluations, allowing one component’s quality to compensate for the other.User-defined weights w_a and w_s determine the influence of the image and service evaluations, with weights summing to 1.
  • Solution selection: The best overall solution is selected by ranking the evaluated feasible image-service pairs.After selection, the image can be deployed, customized, and used to execute the migration strategy.
  • Integrated approach: An integrated MC2 process can select the combination directly, but CloudGenius recommends separate selection because the compensatory relation is less transparent.In the integrated approach, weighting the goal hierarchy controls the trade-off indirectly.
  • Complexity: The approach has complexity proportional to the numbers of images and services, with additional AHP normalization and weight-derivation computations.The computation includes requirement checks, criterion evaluations, feasibility checks, and image-service pair evaluation.

4. EXAMPLE APPLICATION

The example applies CloudGenius to migrating a PHP-based Web shop from local infrastructure to the Cloud. The migration strategy adds the new application URL to an existing load balancer and removes old URLs over time.

  • Example application: An e-business uses CloudGenius after deciding to move its Web shop from costly legacy infrastructure to Cloud infrastructure.The example application is scalable and provides a PHP environment while retaining data in the private data center.
  • Example application: The example’s software stack includes Apache httpd, Perl, PHP, and MySQL.The stack is identified in the example’s supporting note.
  • Migration strategy: The migration strategy adds the new Web application URL to an existing load balancer and removes old URLs over time.Data migration is out of focus in the example.

5. CUMULUSGENIUS PROTOTYPE

CumulusGenius is the software implementation that makes CloudGenius applicable with acceptable effort by implementing its selection model and automating evaluations and deployments.

  • Prototype implementation: CumulusGenius implements models of VM images, services, providers, requirements, and criteria.The implementation addresses the complexity of the selection problem that motivates software support.
  • Prototype implementation: CumulusGenius uses Aotearoa’s API for evaluations and triggers automated test deployments for image-to-service mappings.These components support the practical execution of CloudGenius selection and validation.

6. EXPERIMENTS

Experiments measured CumulusGenius time complexity as the numbers of VM images and services increased. The results show non-linear, quadratic growth and motivate heuristics for large parameter sets.

  • Experimental setup: The experiments compare measured CumulusGenius time complexity with the theoretical complexity of the approach.The evaluation increases images and services from 100 to 1000 in increments of 100 and uses repeated runs.
  • Combination complexity: Combined image-service generation shows similar influence from images and services, but computation effort grows non-linear.The comparison uses measured averages for combinations across the tested input sizes.
  • Total complexity: The aggregated best-combination computation has non-linear growth and is strongly influenced by the number of services.The total values aggregate separate image and service evaluations, grouped combinations, and an overall value.
  • Results and implications: The results confirm quadratic time complexity and show that searching for the best solution is exhaustive as the number of criteria, images, and services grows.The experiments therefore identify scalability as a concern for large parameter sets.
  • Results and implications: Heuristics, including a genetic algorithm using function b as a fitness function, are proposed as an alternative for large parameter sets.This is presented as a first notion for addressing exhaustive search.

7. DISCUSSION & FUTURE WORK

CloudGenius is applicable to single-tier Web application migration but has limited scope and relies on quantitative criteria and full-featured VM images. The framework remains open to broader evaluations and enhancements, including richer databases, benchmarking, and multi-tier support.

  • CloudGenius is limited primarily to single-tier Web applications and does not directly address multi-tier systems with component and persistence-tier dependencies.Multi-tier decision support would need to model relations among components and a persistence tier with distinct characteristics.
  • The framework restricts criteria to quantitative numerical measures because pair-wise evaluation of qualitative criteria across all services and VM images would be immense.
  • CloudGenius assumes that engineers prefer full-featured VM images over basic images requiring customization effort.
  • The framework remains applicable while leaving room for deeper evaluations and enhancements.Suggested extensions include comprehensive, current databases; automated benchmarking; additional cost-calculation tools; richer requirement checks; and support for different image-preparation levels.

8. CONCLUSION

CloudGenius provides a cyclic decision-support process that recommends Cloud VM images and infrastructure services for Web application migration according to engineer requirements and goals. A prototype was validated experimentally, and the underlying algorithm's time complexity was analyzed.

  • CloudGenius provides a cyclic migration process that suggests Cloud VM images and infrastructure services according to an engineer's requirements and goals.
  • The framework addresses non-trivial Web application migration decisions by providing decision support intended to decrease effort and remove hurdles.
  • A prototype implementation was validated in experiments, and the time complexity of its underlying algorithm was analyzed.
Loading 1203.3997v1…