Inhalt des Dokuments
Es gibt keine deutsche Übersetzung dieser Webseite.
|Kick-off meeting [Slides
]||April 10, 2019,
16:15h||H3003A||Nedelkoski , Geldenhuys
|Distr. Stream Processing Intro [Slides ]||April 17, 2019, 16:15h||H3003A||Thamsen|
- This project is limited to 12 students.
- The participation requires a registration.
- To register, please write an email to Sasho Nedelkoski (contact on the right).
- The final list of participants will be determined in the kick-off meeting.
Low-latency processing of large data streams from distributed sensors is becoming increasingly important for a growing number of IoT applications. In these environments sensor data collected at the edge of the network is typically transmitted in a number of hops: from sensor-equipped devices to intermediate resources to large clusters of cloud resources. Deploying tasks of distributed stream processing applications (such as Storm or Flink jobs) across the available resources of these environments can significantly reduce application latencies and network congestion. However, such deployments need to take the heterogeneity of resources and network topologies into account in order to achieve optimal performance. What is more, IoT environments are also inherently dynamic. For instance, the load of a task deployed on a particular resource and the quality of a wireless connection might vary in response to real-word events and changes. Therefore, distributed IoT applications often also need to be resilient against variations in the underlying infrastructures, including tolerating some failures.
During this project, the participants will build an experimental testbed of geo-distributed, heterogeneous resources and will then investigate reliable and efficient designs for distributed processing applications as well as explore possibilities of automatically adapting resource management (e.g. task scheduling utilizing monitoring data and machine learning). Prototypes will be evaluated in the developed testbed environment using a set of IoT application benchmarks.
If you are interested in this project, please write an email until the first meeting (see above) to Sasho Nedelkoski (email on the right side of this page).
Master students with advanced programming skills, who would like to gain practical experience in working on a larger project in the area of distributed and complex IT systems.
- First and foremost, we expect participants to be interested in distributed systems, scalable data processing, and applications of machine learning.
- Then the project will require solid knowledge of working with Linux machines/servers, while some knowledge of Cloud systems and basic understanding of containerized services will be beneficial.
- Finally, good knowledge of at least one programming language is required, while the languages used in this project will likely be Java/Scala and Python.
+49 30 314-24813
Raum TEL 1209
+49 30 314-79675
Raum TEL 1211
+49 30 314-24539