| Objective of Meeting
Summary
Outputs
Next Steps
Links
Sponsoring Forum(s):
Quality of Service
Enterprise Management
|
Meeting Report
QoS Strategy and Standardization
Cannes, France - Tuesday, October 16 2002
Objective of Meeting
The objective of the QoS Strategy and Standardization effort is to
arrive at a standard approach to the propagation of customer-to-vendor
and vendor-to-vendor QoS requirements and measurements in a manner that
is quantifiable, observable, and interoperable, and realizes a process
for end-to-end Quality of Service assurance which is acceptable to
vendors and customers alike.
The strategy is to determine what should be done to make existing QoS
standards, Service Level Agreements, and Policies more effective, and
where standards and policies do not exist, what should be done to create
them
The expected outcome of this session is that participants will have a
better understanding of the goals and objectives of the QoS Task Force
and that the participants will help define action items, milestones and
issues in progressing each of the major deliverables:
Summary
Attendance
Introduction
Jean Hammond began this session by explaining the role of the
SIB, and the purpose of this exercise, to add to The Open Group's
Standards INformation Base ( http://www.opengroup.org/sib/
) those standards that are relevant to achieving the objectives of QoS.
The potential areas for standardization from Carl Bunje's
presentation were reviewed:
|
SLA Specification
- Languages and tools for creating and interpreting SLAs
|
High |
|
Prioritization of Resources
- CPU Resource monitoring and control
- Network traffic differentiatrion and prioritization
- Mechanisms to pass application prioritization and
classifications through OS
|
Medium: need to review what others are
doing in this area. |
Instrumentation and data collection
- Consistent application performance instrumentation
- Metrics at and below middleware layer
- Mechanisms for collecting and labelling contextual/situational
information for performance and failure data
- Machanisms for tying gathered data to application tranmsaction
flow
|
Medium - Low |
Identification of performance bottlenecks
and failures
- Tools for correlation of performance and diagnostic
information across multiple platforms
- Tools which display end-to end views of performance, rather
than component-based approach
- Cross-platform and cross-resource monitoring tools
|
High for Application Manageability |
Automation
- Automated collection and reduction of performance, failure and
contextual data
- Automated mechanisms for prioritized resource reassignment for
service restoration
|
Medium for Resource Allocation |
Token Bucket Regulation
Token Bucket Regulation (TBR) has become a standard part of the open
source packages, Kame and AltQ, which are commonly used in QoS systems.
This presentation addresses scaling issues with VOIP occurring as both
the number of calls and the variance of network QoS increase across a
wide area distribution. The proposed solution is to add a second level
of TBR specifically tailored to streaming data and complementary to the
current TBR implementation.
Charles Richmond presented his work, and a pdf version of the slides
is here.
The object of the exercise is to spread the loss of packets so that,
in applications such as VOIP, the effect of the loss is minimized.
Outputs
The discussion above.
Next Steps
See actions above.
Links
See above.
|
