PhDInformatics Engineering
| Study location | Lithuania, Vilnius, On Campus |
|---|---|
| Academic field | Computer science (ISCED 481) Information systems (JACS I200) |
| Type | Doctoral, full-time |
| Nominal duration | 4 years (30 ECTS) |
| Study language | English |
| Awards | PhD (PhD candidate position in Self-Adaptive QoS-aware container autoscaling using machine learning techniques) |
| Course code | Informatics Engineering T 007 |
| Application fee | €100 one-time |
|---|
| Entry qualification | Postgraduate diploma (or higher) The entry qualification documents are accepted in the following languages: English. Often you can get a suitable transcript from your school. If this is not the case, you will need official translations along with verified copies of the original. You must take verified copies of the entry qualification documents along with you when you finally go to the university. |
|---|
| Language requirements | English International applicants to whom English is not a native language need to provide a proof of their English language proficiency. Exceptions are made only for applicants who have completed their previous studies fully in English. One of the following is accepted: |
|---|
| Other requirements | At least 2 reference(s) must be provided. A relevant portfolio is required. Please upload your research proposal including the abstract, literature review, research objectives, research questions, methodology and bibliography. - Certified copies of the Master’s degree diploma and its supplement with grades or higher education equivalent to it; |
|---|
| More information |
|---|
Overview
PhD candidate position in
Self-Adaptive QoS-aware container autoscaling using machine learning techniques
Open application for a doctoral candidate in Informatics Engineering
The doctoral candidate position is available in Vilnius Gediminas Technical University Faculty of Fundamental Sciences Department of Information Systems. Applicants interested in other research topics related to Informatics Engineering are also welcome to apply.
Research topic description
A wide adaption of containerization technology led to more agile application development, optimization of resource usage, and faster resource provisioning. As applications grow and become more complex, all these instances need to be dynamically scaled up on time in the corresponding amount. Autoscaling service is used for this purpose employing different horizontal and vertical scaling algorithms.
A wide variety of container autoscaling solutions mainly address resource versus cost optimization problems. Also, the Quality of Service (QoS) versus optimization of resource utilization is investigated to minimize the risk of Service Level Agreement (SLA) violations.
Problems arise when a particular Service level objective SLO value based on application performance or user experience must be achieved. The problem can become even worse when the Kubernetes cluster runs on a shared infrastructure that is a public cloud, where the “noisy neighbour” problem arises. Also, autoscaling algorithms do not consider the dynamics of the load, and as a result, resource provisioning is delayed. A slow reaction or latency in the provisioning of required resources causes a QoS drop. To overcome the latency problem an approach of using Machine Learning techniques to predict SLA violations as well as prevent them by means of optimization must be performed.
The research aims to propose and investigate a self-adaptive autoscaling solution and framework for SLA-sensitive applications that will help with service level objectives (SLO) recovery in cases of service level degradation related to insufficient or delayed resource provisioning. Also, a platform that implements the proposed autoscaling algorithms must be developed and experimental research must be performed to evaluate and compare the obtained results with the container autoscaling algorithms proposed by other authors.
The selected candidate will work on the PhD thesis under the supervision of Prof. dr. Dalius Mažeika The successful applicant will have to attend scientific conferences, meetings and internships at other universities.
Requirements
• Required background: Master’s degree in Computer Science
• Expected skills and knowledge:
• Programming experience in one the technology: .NET, Java, Python, JavaScript
• Experience in cloud computing and Kubernetes
• Knowledge of Machine learning and Deep learning
It is a prerequisite you can be present at and accessible to the institution daily.
For more information
Shortlisted candidates will be invited for an interview. The position may not be opened if no qualified candidate is found. Additional information regarding the post may be obtained from Prof. dr. Dalius Mažeika, e-mail: Dalius.Mazeika@vilniustech.lt
Programme structure
The PhD programme consists of:
· Independent research under supervision;
· Courses for PhD students (approximately 30 ECTS credits);
· Participation in research networks, including placements at other, primarily foreign, research institutions;
· Teaching or another form of knowledge dissemination, which is related to the PhD topic when possible;
· The completion of a PhD thesis.