Ένα προσαρμοστικό σύστημα επεξεργασίας σύνθετων ερωτημάτων πάνω στο σύστημα STORM
(EL)
An adaptive complex event processing system over Storm
(EN)
Προυντζος Γεωργιος
(EL)
Prountzos Georgios
(EN)
Δεληγιαννακης Αντωνιος
(EL)
Σαμολαδας Βασιλης
(EL)
Πολυτεχνείο Κρήτης
(EL)
Μανικακη Βασιλικη
(EL)
Γαροφαλακης Μινως
(EL)
Samoladas Vasilis
(EN)
Technical University of Crete
(EN)
Garofalakis Minos
(EN)
Manikaki Vasiliki
(EN)
Deligiannakis Antonios
(EN)
Distributed complex event processing is a method of tracking, analyzing,
processing and detecting specific events or patterns of events that may occur
in event data streams from various distributed sources. Every node within a
network does in-situ processing from its sources. This way of handling large
sets of data whose sources may be geographically scattered, has a couple of
advantages over centralizing the data into a single node and process them
there. The amount of information is potentially vast and limitations of the
available bandwidth render this approach impractical. Thus, a distributed
approach is implemented for communication efficiency. Additionally, having
a single node do all the processing and synchronizations of the network
creates a Single Point of Failure (SPOF) making the system unreliable when
real-time processing is important.
For these reasons, an architecture of an in situ complex event processing
is developed using the Apache Storm primitives. The architecture detects
patterns of interest on incoming event data streams over a number of nodes
in the network in real-time. The patterns, events and network agents are
specified from a user in a file and the system is designed to support multiple
such queries as well as to adapt to any significant changes in the network.
(EN)