The programmable network architectures that emerged in the last decade have opened new ways to enable Autonomic Networks. However, there are several open issues to address before making such a possibility into a feasible reality. For instance, defining network goals, translating them into network
rules, and granting the correct functioning of the network control loop in a self-adaptive manner are examples of complex tasks required to enable an autonomic networking environment. Fortunately, architectures based on the concept of Models at Runtime (MART) provide ways to overcome such complexity. This paper proposes a MART-based framework – using the RFC 7575 as reference (i.e., definitions and design goals for autonomic networking) – to implement autonomic management into a programmable network. The evaluation shows the proposed framework is suitable for satisfying the functional and performance requirements of a simulated network.
GraphPipe - Blazingly Fast Machine Learning Inference by Vish AbramsOracle Developers
GraphPipe is an open source protocol and collection of software designed to simplify machine learning model deployment and decouple it fromframework-specific model implementations.
GraphPipe - Blazingly Fast Machine Learning Inference by Vish AbramsOracle Developers
GraphPipe is an open source protocol and collection of software designed to simplify machine learning model deployment and decouple it fromframework-specific model implementations.
As presented to the Milwaukee Alt.Net group on November 21st, 2011.
UPDATE April 19, 2012: added some domain logic organization slides using Fowler's 4 basic patterns.
Apache® Spark™ MLlib 2.x: migrating ML workloads to DataFramesDatabricks
In the Apache Spark 2.x releases, Machine Learning (ML) is focusing on DataFrame-based APIs. This webinar is aimed at helping users take full advantage of the new APIs. Topics will include migrating workloads from RDDs to DataFrames, ML persistence for saving and loading models, and the roadmap ahead.
Migrating ML workloads to use Spark DataFrames and Datasets allows users to benefit from simpler APIs, plus speed and scalability improvements. As the DataFrame/Dataset API becomes the primary API for data in Spark, this migration will become increasingly important to MLlib users, especially for integrating ML with the rest of Spark data processing workloads. We will give a tutorial covering best practices and some of the immediate and future benefits to expect.
ML persistence is one of the biggest improvements in the DataFrame-based API. With Spark 2.0, almost all ML algorithms can be saved and loaded, even across languages. ML persistence dramatically simplifies collaborating across teams and moving ML models to production. We will demonstrate how to use persistence, and we will discuss a few existing issues and workarounds.
At the end of the webinar, we will discuss major roadmap items. These include API coverage, major speed and scalability improvements to certain algorithms, and integration with structured streaming.
Laravel is a free , open-source PHP web framework created by Taylor Otwell, as an attempt to provide an advanced alternative to CodeIgniter. It was released in June 2011, as Beta version, followed by Laravel 1 in the same month. Since then it has grown up swiftly dominating the industry as developer's first choice.
Visit Here for more information http://www.endivesoftware.com/laravel-development-services.php
PHP frameworks provide developers with a ready-made set of tools to build high-quality web applications quickly and efficiently. Know the Top 10 PHP Frameworks in 2024 that developers must use to streamline development processes and enhance app quality.
The PHP framework is a library that comes with commonly used functions. Since developers don’t have to code these functions from scratch, it saves time.
Now, let’s come to microframeworks. What's the need for a microframework? Can't you use a full-stack framework?
Over the past few years, full-stack frameworks have added new features to build large, complex websites. As a result, it became too hard to make a simple website without all the overheads that come with it.
That's where micro-frameworks came into existence. These microframeworks are stripped-down versions of full-stack frameworks for specific use cases.
In simple terms, consider a full-stack PHP framework like an SUV and a micro-framework like a bicycle. SUVs can accommodate more people but will also use more fuel. At the same time, bikes will have the advantage of being lightweight and will serve the purpose at a low cost.
Now that you know the fundamental difference between PHP frameworks and micro-frameworks, let's check out the top 10 popular frameworks that can fuel your PHP projects.
A gentle introduction to Apache Spark from the theorem of Resilient Distributed Datasets to deploying software to the core platform, Spark Streaming, and Spark SQL
Introduction to YANG data models and their use in OpenDaylight: an overviewCisco DevNet
A session in the DevNet Zone at Cisco Live, Berlin. YANG is a data modeling language defining device and service configuration and operations. This session will describe what YANG is (with examples), and its relationship to and how it is used in OpenDaylight. Several tools making it easier for the application developer to work with YANG will be discussed. It will conclude with a demonstration of YANGUI and YANG Visualizer, two new OpenDaylight applications that auto-generate a user interface and directed graph respectively, both based on selected YANG models.
As presented to the Milwaukee Alt.Net group on November 21st, 2011.
UPDATE April 19, 2012: added some domain logic organization slides using Fowler's 4 basic patterns.
Apache® Spark™ MLlib 2.x: migrating ML workloads to DataFramesDatabricks
In the Apache Spark 2.x releases, Machine Learning (ML) is focusing on DataFrame-based APIs. This webinar is aimed at helping users take full advantage of the new APIs. Topics will include migrating workloads from RDDs to DataFrames, ML persistence for saving and loading models, and the roadmap ahead.
Migrating ML workloads to use Spark DataFrames and Datasets allows users to benefit from simpler APIs, plus speed and scalability improvements. As the DataFrame/Dataset API becomes the primary API for data in Spark, this migration will become increasingly important to MLlib users, especially for integrating ML with the rest of Spark data processing workloads. We will give a tutorial covering best practices and some of the immediate and future benefits to expect.
ML persistence is one of the biggest improvements in the DataFrame-based API. With Spark 2.0, almost all ML algorithms can be saved and loaded, even across languages. ML persistence dramatically simplifies collaborating across teams and moving ML models to production. We will demonstrate how to use persistence, and we will discuss a few existing issues and workarounds.
At the end of the webinar, we will discuss major roadmap items. These include API coverage, major speed and scalability improvements to certain algorithms, and integration with structured streaming.
Laravel is a free , open-source PHP web framework created by Taylor Otwell, as an attempt to provide an advanced alternative to CodeIgniter. It was released in June 2011, as Beta version, followed by Laravel 1 in the same month. Since then it has grown up swiftly dominating the industry as developer's first choice.
Visit Here for more information http://www.endivesoftware.com/laravel-development-services.php
PHP frameworks provide developers with a ready-made set of tools to build high-quality web applications quickly and efficiently. Know the Top 10 PHP Frameworks in 2024 that developers must use to streamline development processes and enhance app quality.
The PHP framework is a library that comes with commonly used functions. Since developers don’t have to code these functions from scratch, it saves time.
Now, let’s come to microframeworks. What's the need for a microframework? Can't you use a full-stack framework?
Over the past few years, full-stack frameworks have added new features to build large, complex websites. As a result, it became too hard to make a simple website without all the overheads that come with it.
That's where micro-frameworks came into existence. These microframeworks are stripped-down versions of full-stack frameworks for specific use cases.
In simple terms, consider a full-stack PHP framework like an SUV and a micro-framework like a bicycle. SUVs can accommodate more people but will also use more fuel. At the same time, bikes will have the advantage of being lightweight and will serve the purpose at a low cost.
Now that you know the fundamental difference between PHP frameworks and micro-frameworks, let's check out the top 10 popular frameworks that can fuel your PHP projects.
A gentle introduction to Apache Spark from the theorem of Resilient Distributed Datasets to deploying software to the core platform, Spark Streaming, and Spark SQL
Introduction to YANG data models and their use in OpenDaylight: an overviewCisco DevNet
A session in the DevNet Zone at Cisco Live, Berlin. YANG is a data modeling language defining device and service configuration and operations. This session will describe what YANG is (with examples), and its relationship to and how it is used in OpenDaylight. Several tools making it easier for the application developer to work with YANG will be discussed. It will conclude with a demonstration of YANGUI and YANG Visualizer, two new OpenDaylight applications that auto-generate a user interface and directed graph respectively, both based on selected YANG models.
Similar to Using the RFC 7575 and Models at Runtime for Enabling Autonomic Networking in SDN (20)
Salas, V. (2024) "John of St. Thomas (Poinsot) on the Science of Sacred Theol...Studia Poinsotiana
I Introduction
II Subalternation and Theology
III Theology and Dogmatic Declarations
IV The Mixed Principles of Theology
V Virtual Revelation: The Unity of Theology
VI Theology as a Natural Science
VII Theology’s Certitude
VIII Conclusion
Notes
Bibliography
All the contents are fully attributable to the author, Doctor Victor Salas. Should you wish to get this text republished, get in touch with the author or the editorial committee of the Studia Poinsotiana. Insofar as possible, we will be happy to broker your contact.
(May 29th, 2024) Advancements in Intravital Microscopy- Insights for Preclini...Scintica Instrumentation
Intravital microscopy (IVM) is a powerful tool utilized to study cellular behavior over time and space in vivo. Much of our understanding of cell biology has been accomplished using various in vitro and ex vivo methods; however, these studies do not necessarily reflect the natural dynamics of biological processes. Unlike traditional cell culture or fixed tissue imaging, IVM allows for the ultra-fast high-resolution imaging of cellular processes over time and space and were studied in its natural environment. Real-time visualization of biological processes in the context of an intact organism helps maintain physiological relevance and provide insights into the progression of disease, response to treatments or developmental processes.
In this webinar we give an overview of advanced applications of the IVM system in preclinical research. IVIM technology is a provider of all-in-one intravital microscopy systems and solutions optimized for in vivo imaging of live animal models at sub-micron resolution. The system’s unique features and user-friendly software enables researchers to probe fast dynamic biological processes such as immune cell tracking, cell-cell interaction as well as vascularization and tumor metastasis with exceptional detail. This webinar will also give an overview of IVM being utilized in drug development, offering a view into the intricate interaction between drugs/nanoparticles and tissues in vivo and allows for the evaluation of therapeutic intervention in a variety of tissues and organs. This interdisciplinary collaboration continues to drive the advancements of novel therapeutic strategies.
The ability to recreate computational results with minimal effort and actionable metrics provides a solid foundation for scientific research and software development. When people can replicate an analysis at the touch of a button using open-source software, open data, and methods to assess and compare proposals, it significantly eases verification of results, engagement with a diverse range of contributors, and progress. However, we have yet to fully achieve this; there are still many sociotechnical frictions.
Inspired by David Donoho's vision, this talk aims to revisit the three crucial pillars of frictionless reproducibility (data sharing, code sharing, and competitive challenges) with the perspective of deep software variability.
Our observation is that multiple layers — hardware, operating systems, third-party libraries, software versions, input data, compile-time options, and parameters — are subject to variability that exacerbates frictions but is also essential for achieving robust, generalizable results and fostering innovation. I will first review the literature, providing evidence of how the complex variability interactions across these layers affect qualitative and quantitative software properties, thereby complicating the reproduction and replication of scientific studies in various fields.
I will then present some software engineering and AI techniques that can support the strategic exploration of variability spaces. These include the use of abstractions and models (e.g., feature models), sampling strategies (e.g., uniform, random), cost-effective measurements (e.g., incremental build of software configurations), and dimensionality reduction methods (e.g., transfer learning, feature selection, software debloating).
I will finally argue that deep variability is both the problem and solution of frictionless reproducibility, calling the software science community to develop new methods and tools to manage variability and foster reproducibility in software systems.
Exposé invité Journées Nationales du GDR GPL 2024
DERIVATION OF MODIFIED BERNOULLI EQUATION WITH VISCOUS EFFECTS AND TERMINAL V...Wasswaderrick3
In this book, we use conservation of energy techniques on a fluid element to derive the Modified Bernoulli equation of flow with viscous or friction effects. We derive the general equation of flow/ velocity and then from this we derive the Pouiselle flow equation, the transition flow equation and the turbulent flow equation. In the situations where there are no viscous effects , the equation reduces to the Bernoulli equation. From experimental results, we are able to include other terms in the Bernoulli equation. We also look at cases where pressure gradients exist. We use the Modified Bernoulli equation to derive equations of flow rate for pipes of different cross sectional areas connected together. We also extend our techniques of energy conservation to a sphere falling in a viscous medium under the effect of gravity. We demonstrate Stokes equation of terminal velocity and turbulent flow equation. We look at a way of calculating the time taken for a body to fall in a viscous medium. We also look at the general equation of terminal velocity.
What is greenhouse gasses and how many gasses are there to affect the Earth.moosaasad1975
What are greenhouse gasses how they affect the earth and its environment what is the future of the environment and earth how the weather and the climate effects.
Using the RFC 7575 and Models at Runtime for Enabling Autonomic Networking in SDN
1. Using the RFC 7575 and Models at Runtime
for Enabling Autonomic Networking in SDN
Felipe A. Lopes
Instituto Federal de Alagoas (IFAL)
2. Autonomic Networking
LOPES, F. A., Using the RFC 7575 and Models at Runtime for
Enabling Autonomic Networking in SDN
Information
Processing
Knowledge
Generation
/ Analysis Policy
Processing
Business Goals /
Policies
Configuration
Context
Based on Strassner, John,
Nazim Agoulmine, and Elyes
Lehtihet. "Focale: A novel
autonomic networking
architecture." (2006).
3. Autonomic Networking in SDN
LOPES, F. A., Using the RFC 7575 and Models at Runtime for
Enabling Autonomic Networking in SDN
Control Plane
Information
Processing
Knowledge
Generation
/ Analysis Policy
Processing
Business Goals /
Policies
Configuration
Context
Based on Stamou, Adamantia,
et al. "Autonomic handover
management for heterogeneous
networks in a future internet
context: A survey." IEEE
Communications Surveys &
Tutorials 21.4 (2019): 3274-
3297.
Challenges
4. Scope:
RFC 7575
IETF’s Network Management Group (NMRG)
LOPES, F. A., Using the RFC 7575 and Models at Runtime for
Enabling Autonomic Networking in SDN
Guidelines and Reference
Model for implementing
autonomic networks
Objective:
Achieve Self-Management
Self-* properties
Self-configuration
Self-healing
Self-protection
Self-optimizing
Design goals
Coexistence with
tradicional management
Decentralization
Distribution
Simplification of NBI
Abstraction
Autonomic monitoring
...
5 more
5. Autonomic Networking in SDN
LOPES, F. A., Using the RFC 7575 and Models at Runtime for
Enabling Autonomic Networking in SDN
Control Plane
Information
Processing
Knowledge
Generation
/ Analysis Policy
Processing
Business Goals /
Policies
Configuration
Context
Based on Stamou, Adamantia,
et al. "Autonomic handover
management for heterogeneous
networks in a future internet
context: A survey." IEEE
Communications Surveys &
Tutorials 21.4 (2019): 3274-
3297.
Challenges
Opportunities
How to enable
autonomic
networking in
SDN?
6. How to enable autonomic networking in SDN?
LOPES, F. A., Using the RFC 7575 and Models at Runtime for
Enabling Autonomic Networking in SDN
Autonomic
Network
Management
is not a new
area
Control
Plane is
software
Automated
Software
Engineering
Community made
contributions
Models at
Runtime
First
papers
from
2000’s
IETF’s
Community
made
contributions
RFC 7575
7. Models at Runtime (MART)
LOPES, F. A., Using the RFC 7575 and Models at Runtime for
Enabling Autonomic Networking in SDN
Structure
Behavior
Goals
System
The word “Models” of MART
comes from the Model-Driven
Engineering (MDE) discipline.
Objective/Adaptation Models
Learning/Monitoring
Reasoning
Analyzer
Executor
Managed System
Aßmann, Uwe, et al. "A
reference architecture and
roadmap for models@ run.
time systems." Models@ run.
time. Springer, Cham, 2014.
1-18.
Objective Management Layer
Configuration Management Layer
Base Layer
8. Models at Runtime (MART)
LOPES, F. A., Using the RFC 7575 and Models at Runtime for
Enabling Autonomic Networking in SDN
The word “Models” of MART
comes from the Model-Driven
Engineering (MDE) discipline
Objective/Adaptation Models
Learning/Monitoring
Reasoning
Analyzer
Executor
Managed System
Aßmann, Uwe, et al. "A
reference architecture and
roadmap for models@ run.
time systems." Models@ run.
time. Springer, Cham, 2014.
1-18.
Metamodels Concrete Syntax Modeled System
Monitoring
Machine Learning
Algorithms
Generated Code
Runtime Models
Code Templates
9. LOPES, F. A., Using the RFC 7575 and Models at Runtime for
Enabling Autonomic Networking in SDN
Autonomic
Network
Management
is not a new
area
Control
Plane is
software
Automated
Software
Engineering
Community made
contributions
First
papers
from
2000’s
IETF’s
Community
made
contributions
How to enable autonomic networking in SDN?
RFC 7575
How to use RFC 7575 and MART for enabling
Autonomic Networking in SDN?
Autonomic Network
Models at
Runtime
10. RFC 7575 and MART for Autonomic
Networking in SDN
LOPES, F. A., Using the RFC 7575 and Models at Runtime for
Enabling Autonomic Networking in SDN
Proposed MART Architecture
RFC 7575 Design Goals
Self-configuration
Self-healing
Self-optimizing
Abstraction
Autonomic Reporting
Independence of Function and Layer
Full Life-Cycle Support
11. RFC 7575 and MART for Autonomic
Networking in SDN
LOPES, F. A., Using the RFC 7575 and Models at Runtime for
Enabling Autonomic Networking in SDN
Models
Metamodels
12. RFC 7575 and MART for Autonomic
Networking in SDN
LOPES, F. A., Using the RFC 7575 and Models at Runtime for
Enabling Autonomic Networking in SDN
Development and Management Models as inputs
Running code / Network rules as outputs
13. Evaluation
Which RFC 7575’s design goals the proposed MART-based solution
could not achieve? Why?
How suitable is the proposed MART-based solution from a performance
perspective?
LOPES, F. A., Using the RFC 7575 and Models at Runtime for
Enabling Autonomic Networking in SDN
14. Evaluation
Which RFC 7575’s design goals the proposed MART-based solution
could not achieve? Why?
LOPES, F. A., Using the RFC 7575 and Models at Runtime for
Enabling Autonomic Networking in SDN
RFC 7575 Design Goals not achieved Reason
Self-protection Currently, metamodels do not have entities for representing such
goal.
Coexistence with Traditional
Management
The final ANM system supports only intents.
Secure by Default The final ANM system cannot assert the membership of all
componentes as required by the RFC.
Common Autonomic Networking
Infrastructure
Achieving this goal using a unique MART solution would increase the
complexity of metamodels.
15. Evaluation
How suitable is the proposed MART-based solution from a performance
perspective?
LOPES, F. A., Using the RFC 7575 and Models at Runtime for
Enabling Autonomic Networking in SDN
Use case
Metamodels and
algorithms implemented
in Graphical Modeling
Framework (GMF)
Modeling Simulation
https://github.com/felipealencar/mdn
Enable the communication in a network topology
consisting of three nodes connected by four switches,
forming two different paths. All the links have 10Mb
of bandwidth.
16. How suitable is the proposed MART-based solution from a performance
perspective?
LOPES, F. A., Using the RFC 7575 and Models at Runtime for
Enabling Autonomic Networking in SDN
Use case Modeling
Configuration Model Objectives Model
17. How suitable is the proposed MART-based solution from a performance
perspective?
LOPES, F. A., Using the RFC 7575 and Models at Runtime for
Enabling Autonomic Networking in SDN
Simulation
Modeling
Code Generation
Monitoring
Learning
Models as inputs of code templates
Mininet network simulation
18. How suitable is the proposed MART-based solution from a performance
perspective?
LOPES, F. A., Using the RFC 7575 and Models at Runtime for
Enabling Autonomic Networking in SDN
Simulation
Modeling
Select actions and network
monitoring to check if h1 reaches h2.
Learning rate: 0.0001
Activation function: ReLu
19. How suitable is the proposed MART-based solution from a performance
perspective?
LOPES, F. A., Using the RFC 7575 and Models at Runtime for
Enabling Autonomic Networking in SDN
Simulation
Modeling
Introduce congestion traffic to verify
model transformations and
reasoning in the action selection
process
20. Conclusions
LOPES, F. A., Using the RFC 7575 and Models at Runtime for
Enabling Autonomic Networking in SDN
Proposal for enabling
autonomic network
management in SDN
Knowledge combination
High-level modeling for
implementing autonomic
behavior
Evaluation in different use cases
and scenarios
Suitable performance
Some design goals could not be
covered
Proposal for enabling
autonomic network
management in SDN
Knowledge combination
High-level modeling for
implementing autonomic
behavior
Evaluation in different use cases
and scenarios
Proper functioning and suitable
performance
Some design goals could not be
covered
Bom dia!
Meu nome é Felipe Alencar, sou professor do IFAL e irei apresentar o artigo “Using the RFC 7575 and Models at Runtime for Enabling Autonomic Networking in SDN” aceito nesta edição do Workshop Pré-IETF.
Então começando pelo conceito de redes autonômicas, ele é utilizado para definir conjunto de procedimentos, dados e interações que resultam em redes que funcionam baseadas em objetivos e executam tarefas (com pouca ou nenhuma interação de um ser humano) que visam alcançar estes objetivos de forma autonômica. Essa imagem, baseada na arquitetura Focale, proposta por Strassner (2006), ilustra estas interações, os procedimentos e os possíveis dados envolvidos no ciclo autonômico para redes tradicionais.
Com o surgimento do paradigma SDN, a grande diferença é que o conjunto de elementos autonômicos agora podem ser implementados em um plano de controle separado dos elementos de encaminhamento. Porém, existem desafios na definição de objetivos ou políticas de rede, na geração e análise de conhecimento, no processamento das políticas de rede, na configuração automática de elementos de rede e no monitoramento utilizado para gerar a base de conhecimento e oferecer suporte ao atendimento dos objetivos
Neste cenário, o Network Management Group do IETF trabalhou na RFC 7575 que define diretrizes a serem seguidas para que o conceito de redes autonômicas seja alcançado. Essa RFC tem em seu escopo 4 propriedades Self-* e 11 objetivos de design para soluções que busquem prover redes autonômicas, como por exemplo: aumento no nível de abstração, monitoramento autonômico e coexistência com o gerenciamento de rede tradicional.
Considerando que os desafios que existem em cada parte necessária para a realização de redes autonômicas podem também ser encarados como a oportunidade de responder à pergunta que motiva este trabalho: “Como habilitar redes autonômicas em SDN”?
Para responder a questão que motiva este trabalho, devemos ter em mente que: a área de gerenciamento autonômico de redes não é nova. Os primeiros artigos surgiram no início dos anos 2000. O plano de controle, em SDN, é software. E já existem contribuições em comunidades que lida com tornar softwares autonômicos.A comunidade de Engenharia de Software Automatizado têm contribuições nessa linha de autonomicidade, o conceito de Modelos em Tempo de Execução é uma dessas contribuições. O próprio IETF contribuiu com a RFC 7575 e grupos de trabalho nesta mesma linha.
E o que são os Modelos em Tempo de Execução?
Este conceito considera que todo sistema observável possui objetivos, comportamento e uma estrutura. Ao considerar esse sistema como software, é possível representá-lo como modelos que interagem entre si, aprendendo, monitorando, analisando e executando a própria representação do sistema. Na categorização de Assmann, essa representação pode ser feita em três camadas: camada de objetivos, camada de configuração e o camada base, onde ocorre a execução do sistema observado.
Ao concretizar cada representação do sistema, temos: primeiro) metamodelos que são usados para definir formalmente os conceitos do sistema observado ou a ser gerenciado; segundo) uma sintaxe que é usada para definir semanticamente os conceitos do sistema, aumentando o nível de abstração; terceiro) modelos de código que têm o papel de permitir a adaptação do que será executado em acordo com o que está modelado; quarto) técnicas de monitoramento e algoritmos de aprendizagem de máquina que devem ser utilizados para classificar e predizer comportamentos do sistema. Assim, essa concretização resulta no sistema modelado em alto nível transformando-se em código gerado que reflete com precisão os modelos criados.
Compreendendo a RFC 75757 e agora o conceito de MART. A pergunta que surge é: como combinar estes dois conceitos para permitir redes autonômicas em SDN.
Para responder à pergunta específica levantada neste ponto, este trabalho propõe uma arquitetura MART composta por 3 camadas que interagem entre si. São elas: i) a camada de modelo de rede, responsável por abrigar metamodelos e modelos a serem usados na modelagem dos objetivos e da própria rede; ii) a camada de adaptabilidade, que abriga a base de conhecimento e elementos de análise, aprendizado e raciocínio (ou predição) e interage tanto com a camada de modelo de rede quanto com a camada de infraestrutura; iii) camada de infraestrutura que por sinal é a última camada prevista na arquitetura proposta prevê uma infraestrutura SDN padrão.
Os elementos dessas camadas foram definidos visando alcançar os objetivos de design definidos pela RFC 7575.
A Camada de Modelo de Rede é totalmente baseada em metamodelos e sintaxes concretas dos elementos do domínio do sistema. Apesar de nesta parte da apresentação serem mostrados apenas dois metamodelos, é importante observar que é nesta camada onde são especificadas as regras sintáticas que compõe cada metamodelo. Por exemplo, se na linguagem de modelagem final iremos modelar um objetivo, são estes metamodelos que garantem que é possível haver um relacionamento transitivo entre ações e objetivos, e entre ações e fluxos, mas não um relacionamento direto entre objetivos e fluxos. Essas regras garantem a boa formação de modelos e a geração correta de código executável.
A segunda camada, chamada Camada de Adaptabilidade, faz uso dos modelos criados na Camada de Modelo de Rede (que podem sofrer atualizações durante a execução). Estes modelos servem como entrada do algoritmo de aprendizagem por reforço profunda que possui o papel de receber os objetivos que foram modelados e buscar ações que alcancem estes objetivos. Enquanto o objetivo modelado não é alcançado, o algoritmo (que é baseado em implementações da literatura) seleciona aleatoriamente uma das possíveis ações do modelo e seu respectivo parâmetro de rede, calcula a variável de aprendizado antes da execução da ação, e depois da ação executada, monitorando a rede, é calculada a recompensa (ou punição) decorrida desta execução. Essa associação entre ação executada e valor da recompensa é armazenada na base de conhecimento, onde um número X de amostras é coletado para comparar o desconto a ser aplicado na recompensa calculada anteriormente. Esse passo é necessário para conseguir fazer a comparação histórica da execução de ações (isso evita que o algoritmo volte a utilizar regras que não estavam melhorando o parâmetro observado). Restando atualizar a última transição com o desconto aplicado e treinar a rede Q (uma rede neural treinada com valores de qualidade, ou Q value, e ações executadas). E, por fim, realizar a comparação entre o estado do objetivo a ser alcançado, o melhor estado até então observado e o pior estado até então observado).
Vale ressaltar que código executável e regras de rede são sempre gerado a partir dos modelos e do algoritmos de aprendizagem.
A avaliação deste trabalho buscou responder duas perguntas, que estão diretamente relacionadas à motivação original sobre como permitir redes autonômicas em SDN. São elas:
Quais objetivos de design da RFC 7575 a solução proposta baseada em MART não pôde alcançar e por que?
Quão viável é a solução proposta baseada em MART sob uma perspectiva de desempenho?
Para responder à primeira pergunta, uma análise qualitativa funciona foi feita e foi possível observar que dos 11 objetivos de design definidos pela RFC 7575, 4 deles não puderam ser alcançados, além de 3 que foram parcialmente alcançados.
Dos que não puderem ser alcançados temos:- O objetivo de auto-proteção que, apenas por falta de inserções de entidades no metamodelo, não pôde ser atingido;
A coexistência com gerenciamento tradicional, que não é possível dado que o suporte do sistema de gerenciamento gerado é apenas para Intents;
A segurança por padrão que a RFC requer garantias de que os componentes que interagem no sistema de gerenciamento sejam identificáveis, isto ficou fora do escopo da metamodelagem;
E o objetivo de definir uma infraestrutura de rede autonômica padrão, que a RFC define para padronizar a definição de todas as funções autonômicas conhecidas da rede. Isso iria aumentar muito a complexidade dos metamodelos da proposta e também foi deixado fora do escopo.
Para avaliar o desempenho e verificar a sua adequação a possíveis implantações reais, foi realizada a implementação da arquitetura proposta utilizando o framework GMF (um framework para gerar ferramentas computacionais baseadas em modelos de alto nível). Além disso, foi definido um caso de uso para ser modelado, que tem como objetivo simplesmente permitir a comunicação entre uma topologia de rede composta por 3 nós, 4 switches e duas rotas. E, por fim, gerada uma simulação para observar métricas de desempenho e o sistema de gerenciamento de rede autonômico em execução. Vale ressaltar que o código da implementação usada na modelagem está disponível no GitHub.
A etapa de modelagem envolve o uso da ferramenta gerada a partir dos metamodelos. Esse uso refere-se a criação do modelo de configuração da rede e a criação do modelo de objetivos da rede. Considerando as informações do caso de uso, temos no modelo de configuração a modelagem dos três nós, conectados por links e 4 switches formando as duas rotas possíveis de comunicação entre esses nós.
No lado do modelo de objetivos, temos que um dos objetivos modelos é permitir o encaminhamento de fluxos entre o nó h1 e o nó h2.
Da modelagem anterior, ocorre a geração de código – baseada na forma de interação com um controlador SDN (o Ryu no caso dessa avaliação) – e esta geração utiliza as informações do modelo para o código final gerado a ser executado pelo controlador. Nesse ponto, o algoritmo de aprendizagem por reforço profunda também é utilizado e suas variáveis são substituídas por valores presentes no modelo neste processo de geração de código.
Com o código gerado, ocorre a simulação na ferramenta Mininet. Por meio de monitoramento, a simulação retroalimenta os modelos, que podem ser atualizados, a depender das ações escolhidas pelo algoritmo de aprendizagem e do reflexo delas na rede.
O primeiro experimento nesse cenário buscou avaliar se o sistema autonômico era capaz de alcançar o objetivo modelado: permitir a comunicação entre os hosts h1 e h2. Neste gráfico podemos ver a evolução das recompensas conforme o algoritmo começa a utilizar regras de rede que o aproximem do parâmetro estabelecido.
Além do comportamento funcional de um simples encaminhamento de fluxos, adicionamos um objetivo de minimização de atraso no modelo de objetivos para verificar o comportamento do algoritmo nesta situação.
Neste cenário de minimização e atraso, também foi possível observar a auto-adaptação que o algoritmo buscou, executando regras de rede de delegação de fluxo como alternativa para alcançar o segundo objetivo.
Concluindo, a proposta de combinação de áreas distintas visando permitir redes autonômicas em SDN pôde contribuir com uma solução alternativa de alto nível para implementar comportamento autonômico no gerenciamento destas redes, solução esta baseada em diretrizes de uma RFC.Como limitações, o trabalho demonstrou alguns objetivos de design que não puderam ser alcançados e que podem ser abordados em trabalhos futuros, como a compatibilidade com o gerenciamento de rede tradicional. A avaliação da solução proposta em diferentes cenários e casos de uso também é um caminho a ser explorado. Porém, os resultados dos experimentos mostraram um funcionamento correto e um desempenho adequado da solução.