diplomsko delo
Bogdan Petrović (Author), Matjaž Kukar (Mentor)

Abstract

Računalništvu v oblaku je v zadnjem času posvečene veliko pozornosti zaradi njegovega potenciala, saj omogoča prilagodljive rešitve, kot so podatkovne baze ali programska oprema na zahtevo. Med razvojem katere koli vrste aplikacije je potrebno vnaprej premisliti o skalabilnosti, prilagodljivosti in oceniti stroške. V diplomskem delu smo preizkusili različne načine vzpostavitve storitve API za uporabo vnaprej naučenega modela strojnega učenja, pri čemer smo uporabili vsebnike in tehnologije za orkestracijo, testirali različne ponudnike oblačnih rešitev ter jih primerjali med sabo. Poskusili smo tudi oceniti stroške vzpostavitve glede na potrebno infrastrukturo za doseganje zadovoljivega odzivnega časa. Kljub temu da je Kubernetes najpogostejša rešitev za orkestracijo vsebnikov, smo pokazali da je AWS ECS dobra alternativa. Pri oblačni platformi Heroku nimamo veliko fleksibilnosti, vendar je vzpostava zelo enostavna.

Keywords

računalništvo v oblaku;strojno učenje;vsebniki;AWS;računalništvo in informatika;univerzitetni študij;diplomske naloge;

Data

Language: Slovenian
Year of publishing:
Typology: 2.11 - Undergraduate Thesis
Organization: UL FRI - Faculty of Computer and Information Science
Publisher: [B. Petrović]
UDC: 004.85(043.2)
COBISS: 31217411 Link will open in a new window
Views: 876
Downloads: 200
Average score: 0 (0 votes)
Metadata: JSON JSON-RDF JSON-LD TURTLE N-TRIPLES XML RDFA MICRODATA DC-XML DC-RDF RDF

Other data

Secondary language: English
Secondary title: Scalable useage of machine learning models in the cloud
Secondary abstract: In recent times, cloud computing is attracting a lot of attention because of its' services such as servers, databases or software on premise. During the development of an application of any type, scalability, flexibility and cost estimates must be considered in advance. In this diploma thesis, we experimented with different methods for establishing an API service for using a pre-trained machine learning model, whereby we used use containers and technologies for container orchestration, test different cloud providers, and compare them to each other. We tried to estimate the costs of the establishment of a scalable sistem depending of infrastructure needed for achieving a satisfactory response time. Despite Kubernetes being the most often used solution for container orchestration, we have shown that AWS ECS is a good alternative. On Heroku cloud platform, we don't have as much flexibility, however the establishment is very simple.
Secondary keywords: cloud computing;machine learning;containers;AWS;computer and information science;diploma thesis;
Type (COBISS): Bachelor thesis/paper
Study programme: 1000468
Embargo end date (OpenAIRE): 1970-01-01
Thesis comment: Univ. v Ljubljani, Fak. za računalništvo in informatiko
Pages: 58 str.
ID: 12037019