Novell SENTINEL 6.1 SP2, Руководство пользователя, Страница: 443 / 528

Sentinel Architecture 443
no
vd
ocx
(e
n)
7 Ja
nua
ry 201
0
Message Bus
The iSCALE message bus allows for independent scaling of individual components while also
allowing for standards-based integration with external applications. The key to scalability is that,
unlike other distributed software, no two peer components communicate with each other directly.
All components communicate through the message bus, which is capable of moving thousands of
message packets per second.
Leveraging the unique features of the message bus, the high-throughput communication channel can
maximize and sustain a high data throughput rate across the independent components of the system.
Events are compressed and encrypted on the wire for secure and efficient delivery from the edge of
the network or collection points to the hub of the system, where real-time analytics are performed.
The iSCALE message bus employs a variety of queuing services that improve the reliability of the
communication beyond the security and performance aspects of the platform. Using a variety of
transient and durable queues, the system offers unparalleled reliability and fault tolerance. For
instance, important messages in transit are saved (by being queued) in case of a failure in the
communication path. The queued message is delivered to the destination after the system recovers
from the failure state.
Figure A-2
iSCALE Message Bus
Channels
The iSCALE platform employs a data-driven or event-driven model that allows independent scaling
of components for the entire system based on the workload. This provides a flexible deployment
model because each customer’s environment varies: one site can have a large number of devices
with low event volumes; another site can have fewer devices with very high event volumes. The
event densities (that is, the event aggregation and event multiplexing pattern on the wire from the
collection points) are different in these cases and the message bus allows for consistent scaling of
disparate workloads.