| Title | Stud ent(s) | Supervisor | Description |
| Enactment engine to run a composition of functions in a serverless computing environment | 1-2 | Sashko Ristov, Thanh-Phuong Pham | details |
| Large-scale distributed computing within IoT-Edge-Cloud continuum | 1 | Sashko Ristov, Thanh-Phuong Pham | details |
| Data Abstraction to Compose Functions in a serverless computing framework | 1 | Sashko Ristov, Thanh-Phuong Pham | details |
| Dynamic and fault tolerant scheduling of complex applications using volatile instances | 1 | Sashko Ristov, Thanh-Phuong Pham | details |
| Inter-host Orchestration Platform Architecture for Ultra-scale Cloud Applications | 1 | Sashko Ristov, Thanh-Phuong Pham | details |
| Disaster tolerant and cost effective replica placement in content delivery networks | 1 | Sashko Ristov | details |
| Experimente und Datenanalyse für Clouds | 1 | Thomas Fahringer | details |
| Title | Enactment engine to run a composition of functions in a serverless computing environment |
| Number of students | 1 or 2 (Can be joined together with “Data Abstraction to Compose Functions in a serverless computing framework”) |
| Language | English |
| Supervisor | Sashko Ristov, Thanh-Phuong Pham |
| Description | The goal of this master thesis is to develop an enactment engine that will run efficiently an arbitrary composition of functions in a serverless computing environment. Students will explore and learn how to structure and model (e.g. based on XML or JSON) a composition in order to define data and control flow. Students can explore to install, configure and deploy several open source frameworks: e.g. Kubernetes (the container cluster), Docker (Container engine) and Fission and / or OpenWhisk (serverless framework). Students can dive into each software in order to develop advance techniques, such as optimize resource utilization or scheduling algorithms. Additionally, students can explore to extend existing composition management systems, such as AirFlow. |
| Tasks |
|
| Theoretical Skills | Distributed Systems, Cloud Computing, Algorithms, Deviceless Computing, Functions as a Service. |
| Practical Skills | Kubernetes, Docker and Fission or OpenWhisk, Java or Python |
| Additonal Information |
| Title | Large-scale distributed computing within IoT-Edge-Cloud continuum |
| Number of students | 1 |
| Language | English |
| Supervisor | Sashko Ristov, Thanh-Phuong Pham |
| Description | A modern computing system should adapt autonomously and automatic to various dynamic and mobile user workloads. For example, some user requests can be finished in seconds and thus can effort to be sent to the Cloud data center. However, many distributed data sources require a real time data processing (within tens or hundreds of ms), which is less than the WAN networks latency and thus the processing has to be performed near the data sources (in Edge). In this thesis, the students should develop a distributed system that will make an intelligent decision when to offload the computing horizontally (to another Edge device) or to Cloud data center (vertically). The goal of the decision making will be to optimize non-functional conflicting objectives (e.q. maximize user requests, minimize the overall latency) Students will explore use cases with moving users / IoT devices and constant (static) Edge/Fog devices with heterogeneous, but limited resources (communication, computation and storage). In order to achieve higher interoperability and portability, as well as to maximize the revenue, students can exploit a serverless approach. In order to evaluate the system’s scalability, students can explore simulations with iFogSim or EdgeCloudSim. |
| Tasks |
|
| Theoretical Skills | Distributed Systems, Cloud Computing, Optimization, Functions as a Service. |
| Practical Skills | Docker, Kubernetes, Fission / OpenWhisk, Raspberry Pi / Android, Java or Python |
| Additonal Information |
| Title | Data Abstraction to Compose Functions in a serverless computing framework |
| Number of students | 1 or 2 (Can be joined together with “Enactment engine to run a composition of functions in a serverless computing environment”) |
| Language | English |
| Supervisor | Sashko Ristov, Thanh-Phuong Pham |
| Description | Many business applications have constraints and flows within business processes. For example, a company should send a bill or invoice only after they send goods. Another example is “proceed with payment” shall enable two activities – “ship goods” and “inform the customer”. More complex example is a car can be delivered to the customer either after the payment is received or the credit is granted. The goal of this master thesis is to abstract input and output data between composed functions in order to simplify the development and composition. The idea is to define an ordered data collection and using workflow patterns to allow the developer an easy way to define control and data flow between the composition of functions with visual components. From this visualized data and control flow, the system will provide a machine readable language (XML / JSON) that can be later executed in enactment engine. Students can dive into each software in order to develop advance techniques, such as optimize resource utilization or scheduling algorithms, i.e. where and when to run specific functions. |
| Tasks |
|
| Theoretical Skills | Distributed Systems, Cloud Computing, Algorithms, Deviceless Computing, Functions as a Service, Business modelling. |
| Practical Skills | Kubernetes, Docker and Fission or OpenWhisk, Java or Python |
| Additonal Information |
| Title | Dynamic and fault tolerant scheduling of complex applications using volatile instances |
| Number of students | 1 |
| Language | English |
| Supervisor | Sashko Ristov, Thanh-Phuong Pham |
| Description | Public cloud providers offer various on-demand IaaS resources with various capacities, prices and reliability. For example, Amazon offers on-demand instances, which are more reliable than their volatile (spot) instances, but more expensive. Amazon can interrupt and retake spot instances if the biding price is lower than the current market price or due to the lack of available resources. On the other side, spot instances are offered up to 90% of the on-demand price. Accordingly, users are allowed to decide about the trade-off between these parameters. The goal of this thesis is to develop a new algorithm that will schedule execution of many complex applications considering fault tolerance, price and performance. Students will explore the behavior of spot instances and can use our system Askalon to run the complex application on real Cloud’s (e.g. Amazon EC2) using APIs and upgrade our VolatileSim simulator. |
| Tasks |
|
| Theoretical Skills | Distributed Systems, Cloud Computing, Optimizaiton. |
| Practical Skills | Cloud APIs, Java |
| Additonal Information |
| Title | Inter-host Orchestration Platform Architecture for Ultra-scale Cloud Applications |
| Number of students | 1 or 2 |
| Language | English |
| Supervisor | Sashko Ristov, Thanh-Phuong Pham |
| Description | Many memory pages in RAM memory of virtual machines are the same as their operating system, code segment and many parts of the data segment are the same. Cooperation among these virtual machines does not exist; neither between virtual machines hosted on different servers, nor among the servers themselves. If an instruction or data access generates a TLB miss at one node, the same instruction or data access will likely generate a TLB miss in other node of the cluster where the oblivious horizontally-scaled VM-siblings are hosted. The goal of this thesis is to develop an automatic and autonomous memory management over multiple servers (and hosted virtual machines). Students should explore how to read memory accesses in virtual and physical memory from the hypervisor, analyze memory access patterns and page faults. Further on, using Apache Kafka will collect data to a centralized location. Students will learn how to design a predictor, which will improve the memory organization and enable better exploitation of memory patterns. The prediction will process data using Apache Spark. The final goal is to reduce the page fault rate by pre-fetching memory pages on VMs on other hosts. In case a group of two students are interested, you can explore the memory management at the container (application) level, beside the hypervisor level. |
| Tasks |
|
| Theoretical Skills | Operating systems, hypervisors, containers, distributed systems, prediction techniques |
| Practical Skills | Apache Kafka, Apache Spark and / or Apache Storm, Java or Python, XEN / KVM, docker. |
| Additonal Information |
| Title | Disaster tolerant and cost effective replica placement in content delivery networks |
| Number of students | 1 |
| Language | English |
| Supervisor | Sashko Ristov, Thanh-Phuong Pham |
| Description | Fault tolerance is an important feature of a distributed system. A user should not be aware of a failure of a single or several system components. Even more, users should achieve minimum (guaranteed) level of services by a distributed system. Replicating the content is an approach to improve the fault tolerance. This thesis focuses on a content placement problem in a content delivery network of data centers. The goal of the thesis is to design and implement a placement algorithm that will determine the optimal number of replicas of all contents in a distributed system considering to optimize the two conflicting objectives: network utilization and fault tolerance. Students can explore modern content file systems (such as IPFS) and dive into their architecture to improve their fault tolerance and performance. |
| Tasks |
|
| Theoretical Skills | distributed systems, cloud computing, optimization |
| Practical Skills | Java or Python, IPFS. |
| Additonal Information |
| Title | Experimente und Datenanalyse für Clouds |
| Number of students | 1 |
| Language | German |
| Supervisors | Thomas Fahringer |
| Description | Das Ziel dieser Arbeit ist die Durchführung einer Serie von Experimenten, um die Eigenschaften und Fähigkeiten von Cloud Infrastrukturen (z.B. Amazon EC2) zu evaluieren. Es werden dabei zahlreiche Virtual Machine Instanzen (VMs) für kleinere Programme getestet. Dabei werden die Zeiten für die VMs und die Programme gemessen und anschließend ausgewertet. Zu den gemessenen Zeiten gehören: Zeit bis eine VM zugewiesen und gestartet wird, Zeit für die Ausführung der Programme (mit Messung von Speicher und CPU Verbrauch), Zeit um die VM wieder freizugeben, uva. Es werden dabei eine große Zahl von Experimenten gestartet (Script Programm). VMs und Programme müssen vorher instrumentiert werden. Die gemessenen Daten müssen in einer Datenbank abgelegt und dann statistisch ausgewertet und visualisiert werden. Eine Besonderheit ist dabei die Berücksichtigung von Spot Instances, die besonders billig aber vom Cloud Provider jederzeit entzogen werden können. Um solche Spot VMs zu bekommen, muss ein sogenanntes Bieterverfahren implementiert werden. Das Ziel dieser Arbeit ist ein besseres Verständnis von Cloud Ressourcen für verschiedene Programme. Dabei soll der Trade-off zwischen Performance und Kosten genauer untersucht werden. |
| Tasks |
|
| Theoretical skills | einfache Kenntnisse im Bereich der Statistik |
| Practical skills | Script Sprache, Datenbanken, Visualisierung von Daten |
| Additional information |