| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 2.03 MB | Adobe PDF |
Authors
Advisor(s)
Abstract(s)
O aumento de popularidade do processamento e execução de código na Cloud levou ao despertar do interesse pela Functions Framework da Google, sendo o objetivo principal identificar pontos de possível melhoria na plataforma e a sua adaptação de forma a responder à necessidade identificada, tal como a obtenção e análise de resultados com o objetivo de validar a progressão realizada. Como necessidade da Functions Framework da Google Cloud Platform verificou-se que seria possível uma adaptação de forma a promover a utilização de serviços de cache, possibilitando assim o aproveitamente de processamentos prévios das funções para acelerar a resposta a pedidos futuros. Desta forma, foram implementados 3 mecanismos de caching distintos, In-Process, Out-of-Process e Network, respondendo cada um deles a diferentes necessidades e trazendo vantagens distintas entre si. Para a extração e análise de resultados foi utilizado o Apache JMeter, sendo esta uma aplicação open source para a realização de testes de carga e medidas de performance do sistema desenvolvido. O teste envolve a execução de uma função de geração de thumbnails a partir de uma imagem, estando a função em execução na framework. Para este caso uma das métricas definidas e analisadas será o número de pedidos atendidos por segundo até atingir o ponto de saturação. Finalmente, e a partir dos resultados foi possível verificar uma melhoria significativa dos tempos de resposta aos pedidos recorrendo aos mecanismos de caching. Para o caso de estudo, foi também possível compreender as diferenças no processamento de imagens com dimensão pequena, média e grande na ordem dos Kbs aos poucos Mbs.
The increase in popularity of code processing and execution in the Cloud led to the awakening of interest in Google’s Functions Framework, with the main objective being to identify possible improvement points in the platform and its adaptation in order to respond to the identified need, also obtaining and analysing the results in order to validate the progress made. As a need for the Google Cloud Platform Functions Framework, it was found that an adaptation would be possible in order to promote the use of cache services, thus making it possible to take advantage of previous processing of the functions to accelerate the response to future requests. In this way, 3 different caching mechanisms were implemented, In-Process, Out-of-Process and Network, each responding to different needs and bringing different advantages. For the extraction and analysis of results, Apache JMeter was used, which is an open source application for implementing load tests and performance measures of the developed system. The test involves executing a function to generate thumbnails from an image, with the function running in the framework. For this case, one of the metrics defined and analyzed will be the number of requests served per second until reaching the saturation point. Finally, and based on the results, it was possible to verify a significant improvement in the response times to requests using caching mechanisms. For the case study, it was also possible to understand the differences in the processing of images with small, medium and large dimensions in the order of Kbs to a few Mbs.
The increase in popularity of code processing and execution in the Cloud led to the awakening of interest in Google’s Functions Framework, with the main objective being to identify possible improvement points in the platform and its adaptation in order to respond to the identified need, also obtaining and analysing the results in order to validate the progress made. As a need for the Google Cloud Platform Functions Framework, it was found that an adaptation would be possible in order to promote the use of cache services, thus making it possible to take advantage of previous processing of the functions to accelerate the response to future requests. In this way, 3 different caching mechanisms were implemented, In-Process, Out-of-Process and Network, each responding to different needs and bringing different advantages. For the extraction and analysis of results, Apache JMeter was used, which is an open source application for implementing load tests and performance measures of the developed system. The test involves executing a function to generate thumbnails from an image, with the function running in the framework. For this case, one of the metrics defined and analyzed will be the number of requests served per second until reaching the saturation point. Finally, and based on the results, it was possible to verify a significant improvement in the response times to requests using caching mechanisms. For the case study, it was also possible to understand the differences in the processing of images with small, medium and large dimensions in the order of Kbs to a few Mbs.
Description
Dissertion to obtain the Master’s degree in Informatics and Computers Engineering
Keywords
Cloud Functions framework Google Processamento Cache Carga Performance Imagem Função Cloud Functions framework Google Processing Cache Load Performance Image Function
Citation
Rosa, Ivan André Gomes - Reducing execution time in faaS cloud platforms. Lisboa: Instituto Português da Qualidade, 2022. Dissertação de Mestrado