Push-pull-based Monitoring

The push-pull-based approach by Jelasity et al. [1] relies on gossip-based communication to monitor the P2P system. The communication takes places between randomly selected neighbors, which exchange their information. To organize and manage the data exchange between arbitrary peers, the monitoring mechanism divides the time into epochs, which in turn consist of a predefined amount of cycles to calculate the global view of a monitored attribute. In the simulation model, every participating peer measures its attributes at the beginning of a new epoch and periodically sends the changing values to a randomly chosen neighbor during each cycle and in turn receives values from other neighbors. Through the aggregation of the measured attributes at each peer, the values converge to the average at the end of an epoch.

SkyEye

The tree-based approach SkyEye by Graffi et al. [2] calculates an unique ID for each participating node. Using the the lookup-functionality of the underlying DHT, the position in the tree and the parent for that node are identified based on the given ID. Over the created topology, each leaf or inner node, periodically send their attributes towards the root. While a leaf only forwards its locally measured information to its parent, inner nodes in the tree aggregate the received information of their children with their locally measured information, which is then forwarded to the parents. The root, as the last node in this chain, calculates the global statistics of all monitored attributes. This global state information of the system is sent down the tree to every inner node and leaf, leading to a proactive result dissemination.

The resulting tree of SkyEye [2]

References

[1] Jelasity, M., Montresor, A., & Babaoglu, O. (2005). Gossip-based aggregation in large dynamic networks. ACM Transactions on Computer Systems, 23(3), 219–252.

[2] Graffi, K., Stingl, D., Rueckert, J., Kovacevic, A., & Steinmetz, R. (2009). Monitoring and Management of Structured Peer-to-Peer Systems. Proceedings of the Ninth International Conference on Peer-to-Peer Computing, 2009 (pp. 311–320). Seattle: IEEE Computer Society.

A A A | Drucken | Impressum | Sitemap | Suche | Kontakt
Zum SeitenanfangZum Seitenanfang