DISTRIBUTED EVENT-DRIVEN MONITORING MODEL
FOR CLOUD DATACENTERS
Masaryk University, Faculty of Informatics, Botanická 68a, 60200 Brno, Czech RepublicDaniel Tovarňák, Tomáš Pitner
C
C
C
C
1
1
2
3
P
P
P
P
Introduction
References
Research Goals
Distributed Event-driven Monitoring Model
[1] Cretu-Ciocarlie, G. F., Budiu, M., and Goldszmidt, M. Hunting
for problems with artemis. WASL’08, USENIX Association, 2008
[2] Etzion, O., and Niblett, P. Event Processing in Action, 1st ed.
Manning Publications Co., Greenwich, CT, USA, 2010.
[3] Tovarňák, D., and Pitner, T. Towards Multi-Tenant and
Interoperable Monitoring of Virtual Machines in Cloud.
SYNASC’12, 2013
PR
PR
PR PR
publish subscribe
P roducer C onsumerPR ocessing agentS ensor
(Virtualized)
Hardware
Application
Operating
System
Middleware
User
1
2
1
2
3
P
behavior &
state events
complex
events
simple
subscriptions
complex
subscriptions
P
P
P
P
Conclusion and Expected Results
This poster presents the notion of Distributed Event-driven
Monitoring Architecture (DEDMA) and its model. When
compared to existing approaches we expect improvements
in the terms of intrusiveness, network overhead, and
throughput with respect to the number of producers,
consumers, volume, velocity, and variability of monitoring
data.
?
The model is founded on
pattern-based publish-subscribe
interaction and Complex Event
Processing principles [3] with
support for multi-tenancy. The
monitored entities can belong
to any of the main cloud layers.
Methodology
The solution and its respective
components and algorithms will
be evaluated experimentally and
compared to existing approaches.
To be able to achieve this, prototype
implementation will be de-
veloped.
PR
1|2|3 – access
permisions
The volume, velocity, and variability of monitoring data
produced by modern cloud datacenters multiply and the
current approaches are insufficient for online processing
of behavior-related monitoring data.
Aims and Objectives
The goal of our research is to design a Distributed Eventdriven
Monitoring Architecture (DEDMA) following a new
monitoring model with focus on novel mechanisms and
algorithms in the areas of monitoring data generation,
production, distributed collection and processing, as well
as multi-tenant monitoring.
Additionally, when monitoring cloud datacenters
and all its main layers simultaneously,
the monitoring system must also be able to
cope with dynamically changing multi-tenant
environment [2].
When monitoring distributed infrastructure,
huge amounts of monitoring data are typically
produced by multiple distributed producers
spread across many individual computing
nodes. There are reports of monitoring data
rates up to 1 MB/s per node [1].
Scope
In order to determine the
state and behavior of a
particular resource all the
relevant data must be collected,
processed and evaluated.
We are interested in
behavior-related monitoring
data (e.g. logs) in particular.