Database Reference
In-Depth Information
Data producer
LLC Controller
In-transit level
reactive
management
Data
consumer
or sink
Service
Manager
Process
Buffer
Data
transfer
Simulation
Buffer
Data blocks
Data
blocks
Forward
Data
consumer
Application level
proactive management
Coupling
In-transit node
Figure 5.7
Overview of the two-level cooperative QoS management
strategy.
the complementary relationship of an element manager and the limited look-
ahead controller (LLC) within an autonomic element. Each manager monitors
the state of its underlying elements and their execution context, collects and
reports runtime information, and enforces the adaptation actions decided by
the controller. These managers thus augment human-defined rules, which may
be error-prone and incomplete, with mathematically sound models, optimiza-
tion techniques, and runtime information. Specifically, the controllers decide
when and how to adapt the application behavior, and the managers focus on
enforcing these adaptations in a consistent and e
cient manner.
We use the Accord programming system described above to address end-
to-end QoS management and control at two levels shown in Figure 5.7. The
first level in this figure is at the end points using adaptive buffer management
mechanisms and proactive QoS management strategies based on online con-
trol and user-defined policies.
5
,
33
,
34
The second level shown in the figure is at
the in-transit processing nodes, which are resources in the data path between
the source and the destination, using reactive runtime management strategies
and adaptive buffer management mechanisms.
35
,
36
These two levels of man-
agement operate cooperatively to address overall application constraints and
QoS requirements.
QoS management at application end-points
The QoS management strategy
at the application end-points combines model-based LLCs and policy-based
managers with adaptive multithreaded buffer management.
37
The application-
level data streaming service consists of a service manager and an LLC con-
troller. The QoS manager monitors state and execution context, collects and
