TU Wien:Energy-Efficient Distributed Systems VU (Brandic)
Daten[Bearbeiten | Quelltext bearbeiten]
Vortragende | [tiss.person:51423 Brandic, Ivona], [tiss.person:298062 Lujic, Ivan] |
---|---|
ECTS | 3 / 4,5 |
Sprache | English |
Links | tiss:194049 , Homepage |
Masterstudium Data Science |
Inhalt[Bearbeiten | Quelltext bearbeiten]
2022S[Bearbeiten | Quelltext bearbeiten]
8 lectures held by three lecturers:
- Introduction to energy-efficient distributed systems
- Taxonomy on energy-efficient systems
- Autonomic cloud management
- Current forms of distributed systems
- Fuzzy hand-off control
- Code offloading
- Geo-temporal conditions
- Controller design for clouds
+ another lecture where the assignment is introduced
Ablauf[Bearbeiten | Quelltext bearbeiten]
2022S[Bearbeiten | Quelltext bearbeiten]
- Lecture part: oral exam at the end of the semester (see below)
- Exercise part: Assignment (semester project) to be done in groups of 3
Benötigte/Empfehlenswerte Vorkenntnisse[Bearbeiten | Quelltext bearbeiten]
2022S[Bearbeiten | Quelltext bearbeiten]
- Programming skills
- At least basic knowledge about Statistical Learning/Machine Learning
Vortrag[Bearbeiten | Quelltext bearbeiten]
2022S[Bearbeiten | Quelltext bearbeiten]
The majority of slide decks (and thus also the covered material) are a compilation of research papers.
Übungen[Bearbeiten | Quelltext bearbeiten]
2022S[Bearbeiten | Quelltext bearbeiten]
Semester project to be done in groups of 3: Energy-aware scheduling of virtual machines using the GWA-T-12 Bitbrains fastStorage dataset (http://gwa.ewi.tudelft.nl/datasets/gwa-t-12-bitbrains). Students were free to choose their preferred programming language and libraries/frameworks.
Overall tasks:
- Find a suitable technique to predict the CPU utilization of virtual machines (e.g., regression models, time series models, LSTM, ...)
- Implement 2 given algorithms + one own algorithm for energy-aware scheduling of virtual machines
- Write a program to simulate the scheduling of virtual machines on physical machines and compare the energy consumption of the scheduling algorithms
- Summarize your findings in a report.
The progress of the project has to be presented during the semester (2 intermediate presentations) + a final presentation at the end of the semester.
Prüfung, Benotung[Bearbeiten | Quelltext bearbeiten]
2022S[Bearbeiten | Quelltext bearbeiten]
50% oral exam, 50% exercise part
Oral exam is done in slots, 4 students in each slot. 3 questions per student:
- Question about the exercise part
- Question about the lecture part
- Question on a research paper (the research paper is assigned on a per-slot basis ~1 week in advance)
Selection of assigned research papers:
Dauer der Zeugnisausstellung[Bearbeiten | Quelltext bearbeiten]
Semester | Letzte Leistung | Zeugnis | |
---|---|---|---|
2022S | 01.07.2022 | 11.07.2022 |
Zeitaufwand[Bearbeiten | Quelltext bearbeiten]
Within 3 ECTS (given sufficient programming knowledge, background in Statistical Learning/Machine Learning and a decent group)
Unterlagen[Bearbeiten | Quelltext bearbeiten]
noch offen
Tipps[Bearbeiten | Quelltext bearbeiten]
noch offen
Verbesserungsvorschläge / Kritik[Bearbeiten | Quelltext bearbeiten]
2022S[Bearbeiten | Quelltext bearbeiten]
- Depending on the presentation slot you are in, you might not get any feedback on your exercise progress
- There is no feedback on the oral exam - at all (you just get the certificate)
- IMHO, the lecture lacks a common theme (besides the fact that everything is somehow related to distributed systems or energy efficiency).