Analytics at the speed of light: Feasibility and challenges for real time analytics of large datasets in hybrid clouds.

''Real-time services'' is a very challenging topic. Running analytics in real-time when there is an abstract network layer makes things even more complicated. The demand to analyze huge data-sets in real time or in the long term has been increasing over the past decade in many sectors including health-care, general science and various online services with a prime example being the trending MMOG community. Combining the network and computation infrastructure efficiently is a challenge that requires careful planning and deployment. This work extends the work that has been done in the field of cloud computing by incorporating the network infrastructure in the analytics procedure. We follow a threefold approach to the problem using mathematical analysis, simulations and real world experiments. The results have shown that real-time analytics over the network is feasible, despite the lack of QoS provisioning in many cases. The bottleneck of the total procedure is oscillating between the network and the computation part of the system, depending on the available computation and networking infrastructure as well as the time complexity of the algorithms used for the analytics.

Subject

analytics
big data
cloud computing
network
real-time
scheduling
transfer

To reference this document use:

http://resolver.tudelft.nl/uuid:8bd464b1-a7b5-44d2-a871-37f5c3e790e5

Part of collection

Student theses

Document type

master thesis

Rights

Files

PDF

mscThesis.pdf

3.22 MB

Close viewer