U.S. patent application number 12/695734 was filed with the patent office on 2011-07-28 for method and apparatus to provide minimum resource sharing without buffering requests.
Invention is credited to Zsolt Kenesi, Benedek Kovacs, Gabor Nemeth.
Application Number | 20110182176 12/695734 |
Document ID | / |
Family ID | 43903978 |
Filed Date | 2011-07-28 |
United States Patent
Application |
20110182176 |
Kind Code |
A1 |
Kenesi; Zsolt ; et
al. |
July 28, 2011 |
METHOD AND APPARATUS TO PROVIDE MINIMUM RESOURCE SHARING WITHOUT
BUFFERING REQUESTS
Abstract
A throttle device is coupled with a central processing unit in a
node for reducing traffic overload in a Next Generation Network
(NGN). The device is coupled with a basic throttle with different
levels of traffic priority and both are situated between a source
node and a target node for processing traffic. When a traffic offer
is received by the throttle, the throttle device by provisionally
updating the basic throttle priority levels, determines whether to
send the traffic offer on to the source node. If the provisional,
traffic priority level is greater than a new traffic priority level
the traffic is admitted and rejected if the updated priority level
is less than the new traffic priority level.
Inventors: |
Kenesi; Zsolt; (Budapest,
HU) ; Nemeth; Gabor; (Budapest, HU) ; Kovacs;
Benedek; (Budapest, HU) |
Family ID: |
43903978 |
Appl. No.: |
12/695734 |
Filed: |
January 28, 2010 |
Current U.S.
Class: |
370/230 |
Current CPC
Class: |
H04L 47/24 20130101;
H04L 47/10 20130101; H04L 47/11 20130101; H04L 47/215 20130101 |
Class at
Publication: |
370/230 |
International
Class: |
H04L 12/56 20060101
H04L012/56 |
Claims
1. A method of reducing traffic overload in a Next Generation
Network (NGN), wherein traffic includes the flow of admitted
requests, calls, packets and tasks, the method comprising steps of:
utilizing a throttle device, coupled with a basic throttle having
different levels of traffic priority, situated between a source
node and a target node for processing a traffic offer received from
the source node, the throttle device configured to cause a central
processing unit to determine whether to send the traffic offer on
to a node targeted by the source node; provisionally updating the
basic throttle, by the throttle device, as if the traffic offer is
admitted, wherein a new traffic priority level of the basic
throttle is determined; admitting the traffic offer if the
provisionally updated basic throttle priority level is greater than
or equal to the new traffic priority level; and rejecting the
traffic offer if the provisionally updated traffic priority level
is less than the new traffic priority level.
2. The method of claim 1, wherein the basic throttle is a Token
Bucket, the traffic priority level is a watermark level and the
Token Bucket having different watermark levels corresponding to
different traffic priorities.
3. The method of claim 1, wherein determining whether to send the
traffic offer to a target node further comprises; determining a
traffic offer rate; determining a class and priority of the
received traffic offer; and updating the traffic offer class and
the priority of the received traffic offer.
4. The method of claim 1, further comprising adjusting the priority
of the traffic offer according to a predetermined threshold,
wherein if the offer rate is higher than the predetermined
threshold the priority of the traffic offer is decreased and if the
offer rate is the same or lower there is no priority
adjustment.
5. The method of claim 4, wherein the capacity of the basic
throttle is left unchanged if the traffic offer is rejected and
updated if the traffic offer is admitted.
6. The method of claim 4, wherein the relationship
r(t.sub.n)=min{.chi.(t.sub.n)/T(t.sub.n),(T(t.sub.n-1)r(t.sub.n-1)-(t.sub-
.n-t.sub.n-1)r(t.sub.n-1)+.chi.(t.sub.n))/T(t.sub.n) measures the
rate of offered and admitted traffic.
7. A throttle device, coupled with a central processing unit in a
node for reducing traffic overload in a Next Generation Network
(NGN), wherein traffic includes the flow of admitted requests,
calls, packets and tasks, the throttle device comprising: the
throttle device, coupled with a basic throttle having different
levels of traffic priority, being situated between a source node
and a target node for processing a traffic offer received from the
source node, the throttle device configured for determining whether
to send the traffic offer on to a node targeted by the source node;
the throttle device provisionally updating the basic throttle as if
the traffic offer is admitted, wherein a new traffic priority level
of the basic throttle is determined; the basic throttle admitting
the traffic offer if the provisionally updated basic throttle
priority level is greater than or equal to the new traffic priority
level; and rejecting the traffic offer if the provisionally updated
traffic priority level is less than the new traffic priority
level.
8. The throttle device of claim 7, wherein the basic throttle is a
Token Bucket, the traffic priority level is a watermark level and
the Token Bucket having different watermark levels corresponding to
different traffic priorities.
9. The throttle device of claim 7, wherein the throttle device,
when determining whether to send the traffic offer to a target
node, is further configured for: determining a traffic offer rate;
determining a class and priority of the received traffic offer; and
updating the traffic offer class and the priority of the received
traffic offer.
10. The throttle device of claim 7, further configured for
adjusting the priority of the traffic offer according to a
predetermined threshold, wherein if the offer rate is higher than
the predetermined threshold the priority of the traffic offer is
decreased and if the offer rate is the same or lower there is no
priority adjustment.
11. The throttle device of claim 10, wherein the capacity of the
basic throttle is left unchanged if the traffic offer is rejected
and updated if the traffic offer is admitted.
12. The throttle device of claim 11, wherein the relationship
r(t.sub.n)=min{.chi.(t.sub.n)/T(t.sub.n),(T(t.sub.n-1)r(t.sub.n-1)-(t.sub-
.n-t.sub.n-1)r(t.sub.n-1)+.chi.(t.sub.n))/T(t.sub.n) measures the
rate of offered and admitted traffic.
13. A system in a Next Generation Network (NGN) for reducing
traffic overload, wherein traffic includes admitted requests,
calls, packets and tasks, the system comprising: a throttle device
coupled with a central processing unit, the throttle device, being
coupled with a basic throttle having different levels of traffic
priority, and being situated between a source node and a target
node for processing a traffic offer received from the source node,
the throttle device configured for: determining whether to send the
traffic offer on to a node targeted by the source node;
provisionally updating the basic throttle as if the traffic offer
is admitted, wherein a new traffic priority level of the basic
throttle is determined; and the basic throttle admitting the
traffic offer if the provisionally updated basic throttle priority
level is greater than or equal to the new traffic priority level;
and rejecting the traffic offer if the provisionally updated
traffic priority level is less than the new traffic priority
level.
14. The system of claim 13, wherein the basic throttle is a Token
Bucket, the traffic priority level is a watermark level and the
Token Bucket having different watermark levels corresponding to
different traffic priorities.
15. The system of claim 13, wherein the throttle device, when
determining whether to send the traffic offer to a target node, is
further configured for: determining a traffic offer rate;
determining a class and priority of the received traffic offer; and
updating the traffic offer class and the priority of the received
traffic offer.
16. The system of claim 13, wherein the throttle device is further
configured for adjusting the priority of the traffic offer
according to a predetermined threshold, wherein if the offer rate
is higher than the predetermined threshold the priority of the
traffic offer is decreased and if the offer rate is the same or
lower there is no priority adjustment.
17. The system of claim 13, wherein the capacity of the basic
throttle is left unchanged if the traffic offer is rejected and
updated if the traffic offer is admitted.
18. The system of claim 17, wherein the relationship
r(t.sub.n)=min{.chi.(t.sub.n)/T(t.sub.n),(T(t.sub.n-1)r(t.sub.n-1)-(t.sub-
.n-t.sub.n-1)r(t.sub.n-1)+.chi.(t.sub.n))/T(t.sub.n) measures the
rate of offered and admitted traffic.
Description
STATEMENT REGARDING FEDERALLY SPONSORED RESEARCH OR DEVELOPMENT
[0001] NOT APPLICABLE
REFERENCE TO SEQUENCE LISTING, A TABLE, OR A COMPUTER PROGRAM
LISTING COMPACT DISC APPENDIX
[0002] NOT APPLICABLE
BACKGROUND OF THE INVENTION
[0003] The present invention relates to telecom network media
overload. More particularly, and not by way of limitation, the
present invention is directed to a system and method for
controlling signaling overload.
PRIOR ART AND PROBLEMS
[0004] In Next Generation Networks (NGNs), many protocols, e.g.,
H.248.11 (also known as Gateway Control Protocol) are used for
controlling media setup of a call. The protocol messages are
processed on a central processing unit (CPU) of corresponding
nodes.
[0005] Different types of nodes have different signal processing
capacity and some nodes might have significantly higher capacity
than others. Because of that there are scenarios, where signaling
overload caused by the source node in a specified target node has a
high probability of occurring.
[0006] Signaling overload causes system performance degradation
even if the node is able to protect itself by rejecting offers.
External overload control mechanisms have been developed to
restrict in advance (in a source node) the traffic that is offered
to a target node. There are call-gapping algorithms that decide
whether the offer should be sent out to the target. If the desired
maximal offer characteristics are known (determined as part of the
external overload control) decision logic in the source node is
referred as a throttle.
[0007] An external overload control mechanism itself can control
different types of descriptors of traffic flows in a system. For
example, Windows-based solutions control the message turnaround
time with a throttle limiting the number of offers in the system
while other solutions work with restricting a percentage of the
offers compared to the previous period of time. Many others like
H.248.11, control the rate of the offers and use token bucket as a
throttle.
[0008] FIG. 1 depicts external overload control architecture.
Typical requirements on an offer rate limiting throttle in H.248.11
are: [a] the rate of the offers sent out should be limited
according to the rate determined by other parts of the external
overload control; [b] an offer should always be sent out once the
limit is not violated; [c] when only one offer can be sent out and
there are two candidates, always send out the one with the higher
priority; and [d] once different traffic classes are defined (on
the same priority level), and there are candidates not to be sent
out because of the limitations then the offer rate representing the
resource should be split according to some defined agreements
between the traffic classes (Service Level Agreements). (req-3)
[0009] The problem with these requirements is that once they are
put into a real system environment it is very hard to decide
whether they are being met or not. Furthermore, the requirements
might have a different interpretation and even concurring exact
definitions when all of them can not be satisfied at the same
time.
[0010] The rate of offers is not violated on average and according
to the watermark level the maximal peak in traffic; thus, the
maximal violation of the rate is also limited. Setting of a
watermark parameter determines how likely the bucket produces
higher throughput rates for short times (violates [a]) or does not
send out candidates although there would not be rate violation
(violates [b]).
[0011] The rate does handle priorities by applying different
watermarks for different priority levels. Thus, throughput
characteristics are different for priorities; i.e., calls with
higher priority cause higher peaks in traffic (requirement [2]
ok).
[0012] Does not handle traffic classification so can not handle
throughput share type of Service Level Agreement (violates
requirement [d])
[0013] There are methods like Weighted Fair Queuing that queues
offers and thus causes delay in the transmission which solves [d].
It is often required to give a solution without using queues but
providing maximal throughput.
[0014] "Rate Based Call Gapping" is a method based on offer rate
and admission rate measuring that provides a solution for all three
requirements without applying queues but its priority handling is
not straightforward.
[0015] This means that if the parameters are set so that they
affect the priority handling then the behavior still depends on
incoming traffic and there is always a positive probability of a
lower priority call being admitted because of throughput share SLA
priority and a higher priority call may be admitted although the
admittance violates SLA agreements.
[0016] It would be advantageous to have a system and method for
resource sharing without buffering requests that overcomes the
disadvantages of the prior art. The present invention provides such
a system and method.
BRIEF SUMMARY OF THE INVENTION
[0017] The present invention discloses a throttle device for
reducing traffic overload in a Next Generation Network (NGN). In
one aspect, the present invention is directed to a method of
reducing traffic overload in a Next Generation Network (NGN),
wherein traffic includes the flow of admitted requests, calls,
packets and tasks. The method utilizes a throttle device, coupled
with a basic throttle that has different levels of traffic
priority, and is situated between a source node and a target node
for processing traffic offers received from the source node. The
throttle device is configured to determine whether to send the
traffic offer on to a node targeted by the source node. The
throttle device is an extension of the basic throttle device and
provisionally updates the basic throttle as if the traffic offer is
admitted. A new traffic priority level is then determined. The
traffic offer is admitted if the provisionally updated basic
throttle priority level is greater than or equal to the new traffic
priority level. The traffic offer is rejected if the provisionally
updated traffic priority level is less than the new traffic
priority level.
[0018] In another aspect, the present invention is directed to a
throttle device for reducing traffic overload in a Next Generation
Network (NGN). The throttle device is coupled with a basic throttle
that has different levels of traffic priority, being situated
between a source node and a target node for processing a traffic
offer received from the source node. The throttle device is
configured for determining whether to send the traffic offer on to
a node targeted by the source node. The throttle device
provisionally updates the basic throttle as if the traffic offer
has been admitted, wherein a new traffic priority level of the
basic throttle is determined. The basic throttle admits the traffic
offer if the provisionally updated basic throttle priority level is
greater than or equal to the new traffic priority level and rejects
the traffic offer if the provisionally updated traffic priority
level is less than the new traffic priority level.
[0019] In yet another aspect, the present invention is directed to
a system in a Next Generation Network (NGN) for reducing traffic
overload. A throttle device is coupled with a basic throttle that
has different levels of traffic priority and is situated between a
source node and a target node for processing a traffic offer
received from the source node. The throttle device is configured
for determining whether to send the traffic offer on to a node
targeted by the source node. The throttle device provisionally
updates the basic throttle as if the traffic offer is admitted. A
new traffic priority level of the basic throttle is then determined
and the basic throttle admits the traffic offer if the
provisionally updated basic throttle priority level is greater than
or equal to the new traffic priority level. The basic throttle
rejects the traffic offer if the provisionally updated traffic
priority level is less than the new traffic priority level.
BRIEF DESCRIPTION OF THE SEVERAL VIEWS OF THE DRAWING
[0020] In the following section, the invention will be described
with reference to exemplary embodiments illustrated in the figures,
in which:
[0021] FIG. 1 depicts an external overload control
architecture;
[0022] FIG. 2 illustrates a high-level block diagram of an overload
control apparatus in accordance with an embodiment of the present
invention;
[0023] FIG. 3 depicts a high-level view of the throttle extension
and its environment according to an embodiment of the present
invention;
[0024] FIG. 4 illustrates a high-level flow diagram of a process
using the throttle wherein Token Bucket is the basic throttle
according to an embodiment of the present invention; and
[0025] FIG. 5 is a graphical simulation of the present invention
using an extended Token Bucket throttle.
DETAILED DESCRIPTION OF THE INVENTION
[0026] As will be recognized by those skilled in the art, the
innovative concepts described in the present application can be
modified and varied over a wide range of applications. Accordingly,
the scope of patented subject matter should not be limited to any
of the specific exemplary teachings discussed above, but is instead
defined by the following description of the invention.
[0027] FIG. 2 illustrates a high-level block diagram of an overload
control apparatus in accordance with an embodiment of the present
invention. The throttle itself is treated as a target (as a
substitute for the actual target) that actually makes decisions
regarding offers upon admittance of an offer. In other words,
throttle 28 is seen by Source 14 as Target 14. Throttle 28 has its
exact capacity determined and expressed in a rate that is an input
of a decision algorithm.
[0028] To provide minimum share, a basic throttle is required to
handle priorities. Previously, throttles typically had two logical
steps: [0029] 1. Determine the priority and class of the offer and
update the estimators [0030] 2. Make a decision upon the gathered
information. (e.g., Token Bucket compares the bucket size to the
watermark, Rate based Call Gapping compares the provisional
admission rate to a goal function.)
[0031] The throttle mechanism of the present invention comprises
three logical steps: [0032] 1. Determine the priority and class of
the offer and update the estimators [0033] 2. Modify (decrease) the
priority level of the excess offers [0034] 3. Make a decision upon
the gathered information. (e.g., Token Bucket compares the bucket
size to the watermark, Rate based Call Gapping compares the
provisional admission rate to a goal function.)
[0035] A Token Bucket throttle does not measure incoming traffic
rate. The throttle device of the present invention actually extends
the Token Bucket throttle with offer rate measurement and priority
level manipulation and measures offer rates. The throttle device is
typically situated in corresponding nodes of a network where offers
are received and processed in a central processing unit of the
nodes.
[0036] Source 12 generates offers for Target 14 and throttle 22 is
positioned between Source 12 and Target 14, typically being
included in Target 14 node. Throttle 22 includes classification and
measurement module 24, decision module 26 and priority adjustment
module 28.
[0037] An offer is placed so that throttle 22 at time t.sub.n has
to make a decision; to accept or reject the offer. (A previous
offer event occurred at time t.sub.n-1). Next throttle 22 is aware
of priority levels and traffic classes and can determine to which
class the offer belongs, e.g., the offer belongs to traffic class
`i` with priority level `j`.
[0038] An incoming offer rate r.sub.i(t.sub.n) vector is calculated
for all traffic classes using an estimator proposed in
PCT/EP2008059693 (publication in January, 2010). If the incoming
offer rate for the given traffic class is higher than the agreed
share, r.sub.i(t.sub.n)>s.sub.ic(t), the priority of the given
offer has to be decreased from `j` to `new`. Otherwise no
modification is needed and new is to be set to j.
[0039] The provisional bucket size is calculated with any Token
Bucket update equation, e.g.,
\beta(t.sub.--n):=\b(t.sub.--{n-1})-\int.sub.--{t.sub.--{n-1}}
{t.sub.--n}c(t)dt+\nu,
where \nu is the number of Tokens needed to be consumed for the
particular offer arrived at t_n. Then the decision is made by the
original Token Bucket algorithm on the modified offer with a new
watermark W.sub.new, i.e., the call is admitted if
.beta.(t.sub.n)<W.sub.new. The call is admitted if
.beta.(t.sub.n)<W.sub.new. If the offer is rejected, the bucket
size is left unchanged b.sub.i(t.sub.n)=b.sub.i(t.sub.n-1) and if
the offer is admitted, the bucket size is updated to
b.sub.i(t.sub.n)=.beta..sub.i(t.sub.n).
[0040] FIG. 3 depicts a high-level view of the throttle and its
environment according to an embodiment of the present invention.
All unnecessary elements of the overload control model shown in
FIG. 2 are left out and the throttle itself is shown as an
individual functional entity. All requests, calls, packets or tasks
named as offers are provided by a source node to the throttle at
first and the throttle decides whether or not to send them to a
target node. As far as the source node is concerned, the throttle
is the target node.
[0041] Definitions of the throttle/apparatus elements and technical
assumptions for the mathematical model: [0042] 1. The throttle is
maps from offer load point process to the set of actions:
{admission, rejection}. This function is characterized with its
rate (c(t)) and peak (W) capacity similar to a Token Bucket and a
decision strategy. Running the decision strategy does not require
capacity of the type that the throttle is characterized with.
[0043] 2. An offer is the event when the throttle has to decide on
admission or rejection. If an offer is admitted, it cannot be
rejected and if the offer is rejected, it cannot be admitted. The
offer has measurable properties of priority and class. [0044] 3.
The offer load or offered traffic is the flow of offers
characterized by a progressively measurable, but not necessarily
stationary, point process marked with the marks from the mark space
that is the direct product of the set of priorities and classes
(the implication is that the probability of two offer events
occurring at the same time is zero). [0045] 4. The class and
priority sets have finite elements. [0046] 5. The throughput, or
admitted traffic, is the flow of admitted offers (throttle admits
an offer). The flow of admitted offers can be conditioned upon the
whole history of the offer load flow and upon the throttle
parameters.
[0047] These assumptions and definitions are only needed to make
the mathematical discussion clear.
[0048] FIG. 4 illustrates a high-level flow diagram of a process
using the throttle extension with Token Bucket as the basic
throttle, according to an embodiment of the present invention. The
process begins at step 402 when an offer is placed to the basic
throttle at time t.sub.n. Next, the process moves to step 404,
where the throttle identifies the priority and class of the offer.
The process continues in step 406, where the throttle updates the
incoming rate measurement regarding priority and class.
[0049] In step 408, the throttle decreases the priority of the
current offer according to r.sub.i(t.sub.n)<s.sub.ic(t.sub.n)
and in step 410 provisionally updates the Token Bucket size as if
the offer had been admitted. .beta.(t.sub.n) is calculated and if
.beta.(t.sub.n) is greater than or equal to the new watermark as
shown in step 412, the process moves to step 414 and the bucket
size is updated according to b(t.sub.n)=.beta.(t.sub.n).
[0050] On the other hand if .beta.(t.sub.n) is less than the new
watermark, as shown in step 416, the process moves to step 418 and
the bucket size is updated with b(t.sub.n)=b(t.sub.n-1).
[0051] A simple estimator is disclosed to measure the rate of the
offered and the admitted traffic because if an external load
control mechanism provides such information then it is more
effective to make decisions as the external load control
mechanism.
[0052] Possible definitions, i.e., measures of the intensity or
rate of a non stationary point process, follow. Let N(t,t-T) be the
counting process that is k whenever in the time interval [t-T,t)
there was k offers or admitted offers depending on what is to be
measured. Suppose that each admitted offer arrives to the throttle
every t.sub.n times. Example definitions:
[0053] 1) Simple average with fixed measure points:
r(t.sub.n)=N(T.sub.i,T.sub.i-T)/T
[0054] The number of offers is counted in T interval but always
started from specified T.sub.i-T, thus the update of r(t) is
independent of the arrivals, is predefined and the value is taken
as constant for periods of T.
[0055] 2) Sliding average on T interval:
r(t.sub.n)=N(t.sub.n,t.sub.n-T)/T.
[0056] The number of offers is counted in T interval and then
divided by T. The value of r(t) is constant in
[t.sub.n-1,t.sub.n).
[0057] 3) Sliding average on T(t) interval:
r(t.sub.n)=N(t.sub.n,t.sub.n-T(t))/T(t).
[0058] This needs to maintain the history of the process for T(t)
for a period of time, but can cause trouble if T(t) is not bounded
or too big, which is often the case. (If T(t) is a stopping time,
e.g., "the time elapsed from the last N event", that requests to
maintain at least N timers, then it is bounded.)
[0059] 4) Recursive sliding average:
r(t.sub.n)=min{.chi.(t.sub.n)/T(t.sub.n),(T(t.sub.n-1)r(t.sub.n-1)-(t.su-
b.n-t.sub.n-1)r(t.sub.n-1)+.chi.(t.sub.n))/T(t.sub.n)}.chi.(t.sub.n)
can be 1 or 0.)
[0060] This maintains only the previous r(t.sub.n-1) and
T(t.sub.n-1) but often T(t.sub.n-1):=T(t.sub.n)=W/c(t.sub.n) is
good. The variable t=time, W is the highest Watermark level in the
Token Bucket and the times t.sub.n, and t.sub.n-1 stands for the
times the current and previous offer has arrived respectively.
.chi.(t.sub.n) equals 1 if an offer arrived at t.sub.n if zero and
if no offer. If .chi.(t.sub.n) is allowed to take higher values it
can handle marked Poisson processes, e.g., those scenarios when
offers, for example consume a different number of tokens in the
bucket upon admission (the estimation is only to measure the rate
and may be independent from the bucket mechanism).
[0061] All the above definitions are asymptotically unbiased in
limit of their parameter but with different efficiency. The first
and the second (using the simplest definition) are easy to
understand, to implement and often are practical using the setting
T.sub.i=every "T=1" second (that is a counter is checked every
second)
[0062] since the average number of offers is taken at every second.
However, since the parameters are fixed they are unable to follow
more frequent changes in traffic than T. The third has the same
disadvantage but the number of events is fixed.
[0063] In this case the definition to be used is the fourth because
it has the best statistical properties, is easy to compute and the
system has to remember for one former event. It will be further
specified and discussed how to choose T(t.sub.n) and T(t.sub.n-1)
according to the desired characteristics of the admitted
traffic.
[0064] Traffic Measurement with the Token Bucket Bound
[0065] To use the fourth definition T(t.sub.n) and T(t.sub.n-1) has
to be specified. Once the desired admitted traffic has the shape
bound by a Token Bucket, with parameters c(t) for rate and W for
watermark, the best choice is T(t.sub.n-1)=T(t.sub.n)=W/c(t). A
constructed throttle should limit the traffic in the same way.
[0066] The offer rate parameter--denoted by r.sub.i(t.sub.n)--is
measured per traffic class. Any of the above parameters can be used
but the recursive sliding window has the most beneficial
statistical properties with a small complexity.
[0067] Provide Traffic Share SLA and Maximal Throughput
[0068] It is possible that the offer rate of a given class is under
its minimum share. In this case the gap between the offer rate and
the minimum share can be understood as free, unused or as remaining
capacity in the system. Then the following requirement can be
proposed as an extension of the original; if there is free capacity
in the system after the Service Level Agreements are met, the
capacity should be split between traffic that wants to use more
capacity in proportion to the traffic offer rate.
[0069] One of the most important benefits of using the throttle
extension disclosed in the present invention is that the invention
can be configured in such a way that the precedence between the
minimum share requirement and the priority handling become clear.
This is not clear in the solutions based on the Rate Based Call
Gapping.
[0070] Priority Handling with Watermark Settings
[0071] An alternate embodiment that provides a simple way to handle
priorities with using only the Watermark settings and not
manipulating the parameters of the traffic estimators. This
embodiment is based on decreasing priority level of offers. An
example follows that illustrates the type of simple setups that are
possible.
[0072] Suppose original priority levels exist in the system for
normal and emergency calls with assigned watermarks:
W.sub.n<W.sub.e. Two more levels W.sub.nr<W.sub.n,
W.sub.er<W.sub.e are introduced for the reduced priority of
normal and emergency calls from the given class `i` respectively.
Now it is clear that W.sub.nr<W.sub.n<W.sub.e and
W.sub.er<W.sub.e but the relation of W.sub.er to W.sub.n and
even W.sub.nr determines the behavior of the system and can be set
by the user.
[0073] Let W.sub.er<W.sub.nr<W.sub.n<W.sub.e then one
class i offer at a higher rate than its agreed share, and emergency
calls are have higher priority than normal calls (it is doubtful
that any operator requires this setting but is still possible to be
set). On the other hand, if
W.sub.nr<W.sub.er<W.sub.n<W.sub.e, normal priority calls
from classes with an offer rate below the agreed share have higher
priority than emergency calls for classes sending over their agreed
share.
[0074] Let W.sub.nr<W.sub.n<W.sub.er<W.sub.e. In this case
emergency calls from any class are always higher priority than
normal calls regardless of traffic shares. Note that the new
priority levels can be understood as a new dimension for priority.
If the parameter settings are clear the priority dimensions have
clear precedence.
[0075] It can be seen clearly that the admission rate of traffic
Class B is similar to the offer rate of Class B, because its offer
rate is always below its minimum share. All the rejected offers are
from Class A since it offers more traffic than its minimum share.
It is also clear that the throughput is maximized.
[0076] In the first period when there is no overload in the system
all offers are admitted (the admission rate for both Class A and B
are at their offer rates). A short peak period comes when the offer
rates increase. This is because of the Token Bucket
characteristics. Then at the overload part the aggregate throughput
is similar to the Token Bucket maximal throughput i.e. the minimum
share is provided for both Class B and class A while the rest of
the capacity is utilized too.
ADVANTAGES OF THE INVENTION
[0077] The main advantage of the throttle extension is that
different watermarks are applied for different priority levels
thus, the throughput characteristics are different for priorities.
The throttle extension can be attached to any call gapping,
throttling, rate limiting mechanism and it keeps their
characteristics while providing minimum share for traffic classes.
An additional advantage is its simplicity and all that is required,
basically, is to measure incoming offer rates. The present
invention can be mixed and used together with other existing
solutions. The mixture and usage influences how strict the priority
handling of the system will be and how important is the fair
sharing compared to priority handling.
[0078] Abbreviations [0079] CPU Central Processing Unit [0080] GPS
Generalize Processor Sharing [0081] IMS IP Multimedia Subsystem
[0082] NGN Next generation network [0083] TCP Transmission Control
Protocol [0084] SCTP Stream Control Transmission Protocol [0085]
SLA Service Level Agreement [0086] WFQ Weighted Fair Queuing
* * * * *