| Application
Level Bandwidth Management and QoS on the WAN
Case Studies |
| "About
200 M of WAN bandwidth being managed by solutions
provided by us today" |
| |
| |
 |
At VFM we have been working on Application
level QoS solutions with our customers since early
2000.
We have worked with our customers, studied their
specific issues in detail with them and have been
able to customize a solution for the Bandwidth Management
they desired. |
| |
We understand various traffic types,
how they behave while being transported across the
network, the nuances of treating different types
of traffic differently, and the various methods
by which we can work towards achieving the desired
allocation of precious WAN bandwidth to various
contending placations.
The clear benefits demonstrated by our approach
and understanding have resulted in about 200 M of
WAN bandwidth being managed by us at various client
locations today.
|
We
follow a four step management process towards managing
bandwidth :
Classify – "You can Manage what you can
see – and higher the quality of visibility,
better is the ability to manage " is the basis
here. We can automatically identify over 270 different
types of traffic through the powerful identification
capability of Packeteer technology. In addition,
classification is on the basis of any criteria from
Layer 2 to Layer 7 or a combination of these. We
could identify and classify Oracle traffic from
a specific Project Group, or all traffic going into
a set of servers – we work on customizing
the traffic tree to the specific needs and nuances
of the customer's network.
Analyze– Having setup the traffic tree, the
next step is to measure the data, analyze the data.
How is bandwidth being consumed? Which applications
are consuming most bandwidth? What are the response
times of various applications? Is a response time
issue linked to starvation of bandwidth to certain
applications– by the more aggressive traffic
types? Am I using the bandwidth for what I had planned
it for? The answers to these and more such questions
are found in this phase. |
 |
Control – A set of utilisation,
response time behavior baselines are drawn and rules
are applied to the traffic accordingly. A powerful
set of control rules help in doing this. We can
set per application minimum/maximum, per user /
user group minimum / maximum bandwidth, allocate
priorities, not allow certain types of traffic and
Invoke TCP Rate Control as a means of proactively
managing TCP
transmissions.
Report – A comprehensive set of reports –
to provide complete and clear information on the
results of the rules/policies applied. This feedback
is an indicator of the efficacy of the applied policies
and if required the same steps are iterated through
for finer management. |
| Our customers where we have implemented
bandwidth management based on Packeteer technology
are using it for diverse purposes that include : |
- Comprehensive profiling and monitoring of
bandwidth usage at the application level, department
level etc.
|
- Contain certain bursty types of traffic so
that they do not monopolize the entire bandwidth
leading to periods of bandwidth starvation'
for other types of traffic and consequent response
time issues.
|
- Reserve bandwidth for certain types of traffic
like Telent, TN5250, TN3270, Citrix etc. so
that these users get the response times required
off interactive applications.
|
- Accelerate certain specific ftps, to higher
levels of transmission.
|
- Acelerate HTTP to Intranet sites, other sites
important to the business and squeeze the rest
of http traffic – during business hours.
|
- Apply admission control to certain types of
traffic like Realaudio, RealVideo, Chat etc.
|
- Provide higher priority to certain important
users, user groups etc.
|
- Allocate bandwidth between various departments
or Project Groups etc.
|
- Through the patented TCP Rate Control technology
of Packetshaper, reduce retransmissions in the
network, reduce router queues and in effect
increase the efficiency of the network and ROI
on the link.
|
