Browsing by Author "Jakovits, Pelle, juhendaja"
Now showing 1 - 11 of 11
Results Per Page
Sort Options
Item Analysis of Practices for Large Scale Configuration Validation - A Case Study(Tartu Ülikool, 2022) Loide, Kaarel; Jakovits, Pelle, juhendaja; Demidov, Jevgeni, juhendaja; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Arvutiteaduse instituutThe move towards containerisation and microservices architecture by software companies has dramatically increased the complexity of configuring modern software systems. The increase in complexity also brings a rise in misconfigurations that contribute heavily to system failures. The DevOps movement amplifies this problem by insisting that developers should own the whole software development lifecycle. Developers now have to configure complex systems while usually not understanding the impact of their decisions. A way to reduce misconfigurations is to use automatic configuration validation software. This software can be easily misused and hard to keep up to date, causing frustration for the end-users and maintainers. This thesis looks into what principles should organisations follow when using automatic configuration validation on a large scale. A real case of a software company with about 400 developers is analysed. The author brings out the problems identified from the analysis and offers solutions for each. Based on the offered solutions, they lead an effort to improve the existing tooling and processes around automatic configuration validation. After the project’s conclusion, the analysis of the results shows that the efforts successfully reduced the maintenance effort and complaints made about the configuration validation tooling and processes. As a final contribution, the author proposes eight best practices that other organisations could follow to improve their usage of automatic configuration validation.Item Apache MiNiFi efektiivsus servandmetöötluse raamistikuna(Tartu Ülikool, 2020) Koljal, Kaire; Jakovits, Pelle, juhendaja; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Arvutiteaduse instituutWith the development of IoT and the growth of data collected by IoT devices, sending all the collected data straight to cloud is not always the best option. Unfortunately, not all IoT devices have enough resources to process all the data. The goal of this thesis is to evaluate if using Apache MiNiFi on resource constrained devices like Raspberry Pi has any benefits compared to sending all the data to cloud and if using such frameworks on edge devices is useful. The goal of the comparison is to assess the amount of Raspberry’s resources used and the amount of data processed in both scenarios. In addition, the difference in latency from edge device to Apache Spark will be compared.Item DevOps ja selle rakendatavuse uuring Põhja-Eesti Regionaalhaiglas(Tartu Ülikool, 2021) Ojasalu, Siim-Morten; Jakovits, Pelle, juhendaja; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Arvutiteaduse instituutDevOps on veel päris uus võrreldes teiste tarkvaraarendusmeetoditega. Selle loomisel püüti vältida varasemate metoodikate vigu ning sellepärast on DevOps arenenud üheks populaarseimaks ja tõhusaimaks metoodikaks. Seda kasutavad mitmed suured tehnoloogiaettevõtted üle kogu maailma. Töö eesmärgiks on uurida DevOps’i teooreetilist poolt – kirjeldada selle olemust ja kasutust ettevõttes ning seejärel luua võimalik lahendus vastava metoodika integreerimiseks Põhja-Eesti Regionaalhaigla IT-teenistusse. Selleks koostab autor IT-teenistuse praeguse olukorra kohta analüüsi, toob välja hetkel esinevad probleemid ja annab hinnangu, kuidas võiks DevOps tulla kasuks sealses keskkonnas. Lõpetuseks koostab autor plaani, mille abil on samm-sammult võimalik DevOps’i IT-teenistusse integreerida ja arendusmetoodikat muuta.Item Lahendus avaandmete jagamiseks Cumulocity IoT platvormilt(Tartu Ülikool, 2023) Kahu, Märten; Jakovits, Pelle, juhendaja; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Arvutiteaduse instituutCumulocity on IoT seadmete ning nende tekitatud mõõtetulemuste hoidmiseks mõeldud platvorm. TÜ Delta keskus hoiab platvormil oma majaautomaatika andurite poolt tekitatud andmeid. Hetkel ei ole mugav neid andmeid kõigile soovijatele avalikustada, kuna Cumulocity platvormile ligipääsemiseks on vaja eraldi kasutajat ning platvormi haldusrakendus ei ole mõeldud oma kasutajamugavuse poolest tavakasutajatele. Töö eesmärk on luua lahendus, mille abil on võimalik andurite mõõtetulemusi soovitud ajaperioodist alla laadida ning neid avalikustada nii lahenduse siseselt kui ka välisesse andmeportaali. Töö teoreetilises osas kirjeldatakse Cumulocity platvormi, selle rakendusliidest TÜ Delta keskuse Cumulocity andmemudelit. Samuti kirjeldatakse lahenduse arhitektuuri, tehnoloogilisi valikuid ning valmivat lahendust ennast.Item Mapping Solar Potential of Tartu(Tartu Ülikool, 2022) Romashchenko, Bohdan; Jakovits, Pelle, juhendaja; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Arvutiteaduse instituutAs the photovoltaic (PV) systems become more and more popular and developed there also grows the need for tools that would help to estimate the usability and impact of such systems on a larger scale. The use case addressed in this work is estimation of solar energy production on the city scale. This information could help city governments to easier find ways to increase solar energy production and, hence, improve energy efficiency of their city. Such information could be obtained by analyzing the rooftops of the building in the city and, luckily, datasets with detailed city geometry for Estonian cities are publicly available. With regards to photovoltaic systems, the datasets mentioned above can be used as a starting point to analyze the roofs, extracting the attributes such as area, tilt and location for further processing. In this work the author aimed to create a solution that would allow to estimate the amount of solar power produced by Tartu or any other Estonian city present in the dataset given by Estonian Land Board. As a result, author has created a prototype consisting of a processing pipeline that does the analysis of geometry and estimation of solar power, as well as a web app to visualize the results. Considering previous work, there have been similar studies which are either not publicly available, not suitable or outdated. The solution described in this work uses the latest technology and no paid external services. Additionally, it’s designed to work with Estonian data out of the box which was not possible to use with some of the existing tools. Hopefully the developed tool will be further improved and used by local governments to improve the energy efficiency in various cities of Estonia.Item ML-TOSCA: ML pipeline modelling and orchestration using TOSCA(Tartu Ülikool, 2023) Valdas, Artjom; Dehury, Chinmaya Kumar, juhendaja; Jakovits, Pelle, juhendaja; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Arvutiteaduse instituutIn today’s world, machine learning is increasingly involved in different areas. Moreover, automating machine learning workflows through AutoML enables organizations to develop and deploy machine learning solutions at scale rapidly. Additionally, leveraging the power of cloud computing can provide even greater scalability and flexibility, allowing us to efficiently process large datasets and cost-effectively train and implement complex machine learning models. Undoubtedly, these technologies will play an essential role in shaping the future across various industries. Despite many advantages, there is a lack of widespread combined implementations of AutoML and cloud-based solutions. This thesis describes a new AutoML integration approach to the TOSCA standard. TOSCA is an open-source specification used to describe the topology of cloud applications and services. Incorporating AutoML techniques into TOSCA enables users to automatically generate optimized machine learning models with the help of cloud applications, which can improve the speed and efficiency of model creation. The proposed approach is implemented in the RADON ecosystem, allowing node and relationship types to be created. The final solution allows users to create and join blocks to define a complete machine learning pipeline structure.Item Parallel and Cloud-Native Secure Multi-Party Computation(Tartu Ülikool, 2022) Tali, Kert; Talviste, Riivo, juhendaja; Jakovits, Pelle, juhendaja; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Arvutiteaduse instituutSecure multi-party computation (MPC) enables analysis based on sensitive data from multiple data owners, applying distributed cryptographic protocols to ensure privacy. Such protocols introduce distinct communication requirements, causing the computation to run significantly longer than its counterpart, conventional computing. General MPC frameworks are available that make it simple to develop such privacy-preserving applications, but running said applications assumes multiple non-colluding computing parties that host the protocol runtimes, having rigorously set up the required infrastructure. Utilising cloud resources for this occasion is a good alternative to on-premises deployments. First, it allows for a larger degree of automation in the infrastructure set-up. Secondly, cloud datacenters enjoy superior network characteristics, detrimental for MPC performance, and offer elastic compute resources at competitive price models. This thesis presents a cloud-native deployment of the SHAREMIND MPC framework on Kubernetes. It further proposes methods for parallel programming, with which MPC applications could be scaled over clusters. Familiar programming models, MapReduce and bulk-synchronous parallel, are adapted to MPC, and benchmarked in commodity clouds, showing near-linear speedup.Item Real time vs micro-batching in streaming data processing: performance and guidelines(Tartu Ülikool, 2021) Alizada, Mansur; Jakovits, Pelle, juhendaja; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Arvutiteaduse instituutData is used in every second of our life. Nowadays, the majority of this data is coming through the Internet. For providing better fast and scalable service, technologies needed to be efficient and scaled regarding those needs. The initiative of this thesis to provide simple workload for engine comparison. In this master thesis, I will focus on Apache Flink, Spark Streaming, Apache Kafka, Apache Storm, Storm Trident for real-time and micro-batch in streaming data processing. This thesis aims to show the comparisons among those technologies.Item Scheme design in non-relational model database to migrate data from relational model database(Tartu Ülikool, 2020) Garayev, Soltan; Jakovits, Pelle, juhendaja; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Arvutiteaduse instituutThe usual preference of storing data is Relational Databases for enterprise applications. It provides relational model, and helps to keep data in organized structure. For last few decades relational database models have been the first choice of the companies. Things started to change with the term of Big Data which is about storing all kind of data related to end-user behaviour. User behaviour is random which can be understood non-structural as well. In this scenario usual relational database model does not fulfill the requirements. There was a need, and a new concept named NoSQL was the best option to fill this gap. NoSQL means Not Only SQL, and it provides a suitable environemtn to companies to store data in any form without having to define a structure. As time passes non-relational databases became priority in other concepts such as storing not only randomly structured data but structured data as well. If we think about big tech companies like Facebook or Google usual relational database model is unsatisfactory for storing huge amount of data. Companies started to move data from relational database to non-relational databases to store their data. There are articles/papers on data migration from relational to non-relation databases. The main issue here is designing the scheme in non-relational data model based on existing complex relational data model. In this research we aim to suggest ways to solve this issue.Item Synthetic Sensor Data Generation for Authentic IoT Device Emulation(Tartu Ülikool, 2021) Khan, Afsana; Jakovits, Pelle, juhendaja; Adhikari, Mainak, juhendaja; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Arvutiteaduse instituutThe Internet of Things (IoT) is the concept of connecting everyday physical objects to the internet and making them capable of identifying and communicating with other devices. The rise of the Internet of Things in the past few decades is commendable. This significant growth has resulted in IoT becoming a budding industry that provides a variety of products, services, and systems and therefore requires quality assurance. These systems are very complex in nature and are used in real-world environments. For these reasons, IoT systems and devices are obliged to testing. The implicit heterogeneity of IoT systems, which typically consist of many thousands of interacting actuators, sensors, and people, makes this an incredibly challenging task. A very important aspect when testing Internet of Things systems at the system level is to include the behavior exhibited by local entities such as IoT devices. However, physically deploying such a huge number of IoT devices is not always feasible in terms of cost, resource availability, and time. To solve this particular problem, it is essential to make use of data generators for synthetically generating IoT data that can emulate the behavior of real IoT devices. Motivated by this challenge, this thesis focuses on evaluating open source tools for generating real-like IoT data based on existing captured data traces, weighing their pros and cons, and finally proposing an approach that would perform more efficiently to achieve the required goal as well as be reusable. In addition, the proposed approach is evaluated by generating synthetic data for some existing datasets and the results are compared.Item Visualization of rooftop solar potential in smart cities(Tartu Ülikool, 2023) Kolk, Alan; Jakovits, Pelle, juhendaja; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Arvutiteaduse instituutVisualizing 3D building rooftops with solar potential data on a map-based application gives a new look into the possible gains of using rooftop solar panels in a smart city. For local governments, cities, and individuals, visualizing the solar potential of buildings allows for a more informed evaluation of what the maximum gain from using solar panels on a roof would be. In this work, a solution for visualizing 3D city buildings in the 3DCityDB web map client was analysed and developed as a continuation of a previous thesis project. Scripts were created, that directly colour KML 3D building rooftops, based on the rooftops solar potential. The finished solution is made using open-source software and is capable of visualizing Estonian CityGML data attributed with solar potential data. Issues of the previous continuation project were fixed, and points of future developments are discussed. The finished work also serves as a basis for future 3D visualizations using CityGML data and the web application can be further improved to support more types of semantical 3D visualizations.