CN101043457B - Packet wideband monitoring method and its apparatus, packet discarding probability tagging device - Google Patents
Packet wideband monitoring method and its apparatus, packet discarding probability tagging device Download PDFInfo
- Publication number
- CN101043457B CN101043457B CN2007100900433A CN200710090043A CN101043457B CN 101043457 B CN101043457 B CN 101043457B CN 2007100900433 A CN2007100900433 A CN 2007100900433A CN 200710090043 A CN200710090043 A CN 200710090043A CN 101043457 B CN101043457 B CN 101043457B
- Authority
- CN
- China
- Prior art keywords
- message
- token
- probability
- token bucket
- loss priority
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Fee Related
Links
Images
Abstract
The provided method for monitoring message bandwidth comprises: obtaining the current token quantity in the token bucket when labeling PRI for message; according to current token quantity, calculating the probability of abandoning PRI mark of the message; according to the abandoning probability, random labeling the abandoning PRI; then according to last labeling result, controlling the bandwidth occupancy of message. This invention can guarantee fair sharing bandwidth for multiple pieces of flow message.
Description
Technical field
The present invention relates to communication technical field, particularly a kind of packet wideband monitoring method and device thereof, packet discarding probability tagging device.
Background technology
On the network equipment, it is an importance that guarantees the system service quality that the message bandwidth of user's input is supervised.For the bandwidth of user's incoming message is supervised, Committed Access Rate (CAR, Committed Access Rate) is the way of normal employing.The length of user's message is compared in the CAR employing and the token number in the CAR token bucket carries out the loss priority mark to message, for example be red, yellow and green with packet labeling, in the system bandwidth deficiency, abandon the high loss priority of part, for example red, yellow message is realized the purpose that the bandwidth to user's incoming message limits.
There is a kind of timing to drive the packet labeling method of computational token at present, specifically comprises:
Step 10: regularly in token bucket to the token that adds fixed qty;
Wherein, token bucket comprises committed burst yardstick (CBS, Committed Burst Size) token bucket (being called for short the C bucket) and peak value burst yardstick (PBS, Peak Burst Size) token bucket (being called for short the P bucket), the token quantity of adding in the C bucket is committed information rate (CIR, Committed Information Rate), the token quantity of adding in the P bucket is peak information rate (PIR, Peak Information Rate), maximum token quantity is CBS in the C bucket, and maximum token quantity is PBS in the P bucket.
Step 11: message is carried out mark according to the token quantity comparative result in the message length that receives and two token buckets;
If the token quantity that message length needs is greater than the token quantity in the P bucket, message is marked as redness, if but message length is less than the token quantity in the P bucket greater than the token quantity of C bucket, message is marked as yellow, if above two conditions do not satisfy, be that message length is not only less than token quantity in the P bucket but also less than the token quantity in the C bucket, message is marked as green.
Step 12: deduct token quantity and refreshing token bucket in C bucket or the P bucket according to the mark result;
If message is marked as green, C bucket and P bucket are all deducted message length corresponding token quantity, if message is marked as yellow, then the P bucket is deducted message length corresponding token quantity, C bucket token quantity is constant, if message is marked as redness, the token quantity of C bucket and P bucket all remains unchanged.
The packet labeling method logic of above-mentioned timing driving computational token is fairly simple, and the resource that is consumed is fewer, but the wideband monitoring precision is lower, so error can be bigger.
Also have a kind of message to drive the packet labeling method of computational token at present, specifically comprise:
Step 20: according to token quantity time T in the last refreshing token bucket
N-1With current time T
n, calculate the time difference between the two;
C bucket refresh time is poor: Δ t
c=T
n-T
Cn-1
P bucket refresh time is poor: Δ t
p=T
n-T
Pn-1
T wherein
Cn-1Be the last refresh time of C bucket, T
Pn-1It is the last refresh time of P bucket.
Step 21: according to existing token quantity K in the C bucket
Cn-1With existing token quantity K in the P bucket
Pn-1, and two token bucket refresh time difference Δ t
cWith Δ t
p, calculate current token quantity in C bucket and the P bucket respectively;
The current token quantity of C bucket: K
Cn=K
Cn-1+ Δ t
c* CIR
The current token quantity of P bucket: K
Pn=K
Pn-1+ Δ t
p* PIR
Wherein, CIR is the speed of adding token in the C bucket, and PIR is the speed of adding token in the P bucket.
Step 22:, message is carried out mark and deducts token quantity in the corresponding token bucket, the token quantity in the refreshing token bucket according to the current token quantity of C bucket and P bucket.
The packet labeling method that above-mentioned message drives computational token can obtain higher bandwidth supervision precision, but when a plurality of stream shared bandwidth, the message of a plurality of streams exists certain sequencing on the time of carrying out mark, and the token quantity of token bucket and mark time correlation connection, therefore, there is fixing difference in a plurality of stream packets token quantity in the pairing token bucket when carrying out mark, and packet labeling rule foundation is the result that message length and token quantity compare, therefore for the message of each fixing stream of message length, the chance of the token that obtains just has fixing difference, whole messages that can occur some stream when extreme are marked as red situation, therefore can cause the unfairness in the bandwidth supervision.
Summary of the invention
The technical problem that the embodiment of the invention will solve provides a kind of method of packet wideband monitoring and device thereof, packet discarding probability tagging device, can realize that the message of a plurality of streams can be shared the broadband liberally.
The objective of the invention is to be achieved through the following technical solutions:
The embodiment of the invention provides a kind of method of packet wideband monitoring, comprising:
Current token quantity in the token bucket when obtaining message and carrying out priority flag;
According to token quantity current in the described token bucket, the full situation of the sky of computational token bucket expires the loss priority marking probability that situation is calculated message according to the sky of described token bucket again; The full situation of the sky of described token bucket is token quantity current in the described token bucket and the ratio between its burst yardstick; The loss priority marking probability that the full situation of described sky according to described token bucket is calculated message is to calculate by the random labelling probability function between structure random labelling probability and the empty full situation; Loss priority marking probability and a random number of message are compared, message is carried out the loss priority mark according to comparative result;
According to the result of packet loss priority flag, the control message is to the situation that takies of bandwidth.
The embodiment of the invention also provides a kind of device of packet wideband monitoring, comprising:
Token bucket is used for deposits tokens;
Computing unit, the current token quantity of token bucket when being used for calculating message and carrying out priority flag;
The marking probability unit is used for the current token quantity of token bucket that draws according to described computing unit, calculates the loss priority marking probability of message;
Described marking probability unit comprises:
The probability function unit is used to construct the random labelling probability function;
The probability calculation unit is used for the current token quantity of token bucket that random labelling probability function and described computing unit according to described probability function unit structure calculate, and calculates the loss priority marking probability of message;
The packet labeling unit is used for the packet loss priority flag probability that draws according to described marking probability unit, and message is carried out the random drop priority flag;
Described packet labeling unit comprises:
The random number generation unit is used to produce a random number;
Indexing unit is used for the loss priority marking probability of message that described marking probability unit is drawn and random number that described random number generation unit produces and compares, and according to comparative result message is carried out the loss priority mark;
The broadband control unit carries out the result of loss priority mark according to described packet labeling unit to message, and the control message takies situation to the broadband.
The embodiment of the invention also provides a kind of device of packet discarding probability tagging, comprising:
The marking probability unit is used for the token quantity current according to token bucket, calculates the loss priority marking probability of message;
Described marking probability unit comprises:
The probability function unit is used to construct the random labelling probability function;
The probability calculation unit is used for random labelling probability function and the current token quantity of token bucket according to described probability function unit structure, calculates the loss priority marking probability of message;
The packet labeling unit is used for the packet loss priority flag probability that draws according to described probability calculation unit, and message is carried out the random drop priority flag;
Described packet labeling unit comprises:
The random number generation unit is used to produce a random number;
Indexing unit is used for the loss priority marking probability of message that described marking probability unit is drawn and random number that described random number generation unit produces and compares, and message is carried out the mark of random drop priority according to comparative result.
The method of the packet wideband monitoring that provides by the embodiment of the invention and device thereof, packet discarding probability tagging device, adopt the random labelling method that message is carried out the loss priority mark, owing to message is carried out the loss priority mark carries out at random, therefore for the message of the regular length of sharing each stream of broadband, the chance of the token that message obtains of each stream does not have fixing difference, and therefore the message of a plurality of streams can be shared the broadband liberally.
Description of drawings
Fig. 1 is the structure chart of packet wideband monitoring device of the present invention preferred embodiment;
Fig. 2 is the signaling process figure of packet wideband monitoring method preferred embodiment of the present invention;
Fig. 3 is first kind of marking probability curve chart of packet discarding probability tagging method of the present invention;
Fig. 4 is second kind of marking probability curve chart of packet discarding probability tagging method of the present invention;
Fig. 5 the present invention is based on the marking probability curve chart that first kind of marking probability curve chart revised the curve controlled dot generation;
Fig. 6 the present invention is based on first kind of marking probability curve chart to revise the marking probability curve chart that function coefficients generates.
Embodiment
The embodiment of the invention provides a kind of method of packet wideband monitoring and device thereof, packet discarding probability tagging device.
For making technical scheme of the present invention clearer, below with reference to accompanying drawing and enumerate embodiment, the present invention is further elaborated.
See also Fig. 1, be the structure chart of packet wideband monitoring device of the present invention preferred embodiment.Described packet wideband monitoring device comprises token bucket 10, token input unit 11, token bucket refresh unit 12, computing unit 13, counter 14, marking probability unit 15, packet labeling unit 16 and broadband control unit 17.
Described token bucket 10 is used for deposits tokens, and comprises C bucket and P bucket, and the token quantity of depositing in the C bucket is committed burst yardstick (CBS) at most, and the token quantity of depositing in the P bucket is the peak value yardstick (PBS) that happens suddenly at most.
Described token is thrown in unit 11 and throw in the token of some with given pace in the C of described token bucket 10 bucket and P bucket.
Described counter 14 is the timing counter, links to each other with token bucket refresh unit 12, packet labeling unit 15, is used to write down message is carried out the time of loss priority mark and the refresh time of token bucket 10.
In order to reduce resource consumption, counter 14 in the embodiment of the invention can adopt the less counter of bit wide, at this moment, when described packet wideband monitoring device precision setting is higher, counter 14 can overflow by generating period, again count after overflowing, and change token bucket refresh time is that counter 14 overflows the time.
The current token quantity of token bucket when carrying out priority flag that described computing unit 13 is used for calculating message, it comprises time difference calculating unit 131 and token computation unit 132.What described time difference calculating unit 131 was used to calculate described counter 14 records carries out time difference between the time of loss priority mark and time that token bucket refreshes to message, and when described counter 14 generating periods overflow, the time difference between the time of overflowing of described time difference calculating unit 131 computing counters 14 and the last time token bucket refresh time; Remaining token quantity in time difference that described token computation unit 132 calculates according to time difference calculating unit 131 and the token bucket 10 is calculated the C bucket of token bucket 10 when message carried out the loss priority mark and the token quantity in the P bucket respectively.
Described token bucket refresh unit 12 is used for carrying out the result of loss priority mark and the current token quantity of token bucket that computing unit 13 obtains, 10 token quantity in the refreshing token bucket according to the 16 pairs of messages in packet labeling unit.And when counter 14 overflowed, described token bucket refresh unit 12 is the C bucket of refreshing token bucket 10 and the token quantity in the P bucket also.
Described marking probability unit 15 is used for the current token quantity of token bucket that draws according to described computing unit 13, calculates the loss priority marking probability of message, and it comprises probability function unit 151 and probability calculation unit 152.Described probability function unit 151 is used to construct and revise the functional relation of random labelling probability; Random probability function relational expression, the C bucket of computational token bucket 10 and the marking probability of P bucket that described probability calculation unit 152 produces according to probability function unit 151.
Described packet labeling unit 16 is used for the packet loss priority flag probability that draws according to described marking probability unit 15, and message is carried out the random drop priority flag, and it comprises random number generation unit 161 and indexing unit 162.Described random number generation unit 161 is used to produce a random number; Described indexing unit 162 is used for the C bucket of token bucket that probability calculation unit 152 is calculated and the marking probability of P bucket, the random number that produces with random number generation unit 161 compares, according to comparative result message is carried out mark, and the mark result is informed token bucket refresh unit 12.
Described broadband control unit 17 is used for carrying out according to the 16 pairs of messages in described packet labeling unit the result of loss priority mark, and the control message takies situation to the broadband.
See also Fig. 2, be the flow chart of packet wideband monitoring method preferred embodiment of the present invention.Described packet wideband monitoring method comprises:
Step 201: calculate message is carried out difference between time of current time of mark and last refreshing token bucket;
When message carries out priority flag, calculate message carried out difference between time of current time of mark and last refreshing token bucket:
C bucket refresh time is poor: Δ t
c=T
n-T
Cn-1
P bucket refresh time is poor: Δ t
p=T
n-T
Pn-1
T wherein
Cn-1Be the last refresh time of C bucket, T
Pn-1Be the last refresh time of P bucket, T
nCarry out the current time of mark for message.
Step 202: according to original token quantity K in the C bucket
Cn-1With original token quantity K in the P bucket
Pn-1, and two token bucket refresh time difference Δ t
cWith Δ t
p, calculate current token quantity in C bucket and the P bucket respectively;
The current token quantity of C bucket: K
Cn=K
Cn-1+ Δ t
c* CIR
The current token quantity of P bucket: K
Pn=K
Pn-1+ Δ t
p* PIR
Wherein, CIR is the speed of the token quantity of adding in the C bucket, and PIR is the speed of the token quantity of adding in the P bucket.
Step 203: adopt the dynamic random labeling method that message is carried out the loss priority mark;
Described dynamic random labeling method specifically comprises:
Steps A: according to token quantity current in the token bucket, the full situation of the sky of computational token bucket;
Ratio between the burst yardstick that the full situation of the sky of described token bucket is token quantity and token bucket,
Empty full situation: the s_c=K of C bucket
c/ CBS
Empty full situation: the s_p=K of P bucket
p/ PBS
Wherein, K
c, K
pBe respectively current token quantity in C bucket and the P bucket, CBS, PBS are respectively maximum token quantity in C bucket and the P bucket.
Step B: the full situation of the sky of computational token bucket, the loss priority marking probability of calculating message;
Promptly, calculate the marking probability F (s_c) of C bucket and the marking probability F (s_p) of P bucket according to the full situation of sky of described token bucket;
Relation between random labelling probability and the empty full state can adopt first kind of relation function F (s) expression:
Relation function F (s) is a partitioned representation formula, and its variable s represents the ratio between the burst yardstick of token quantity and token bucket.When S_iflx≤s≤S_0, F (s) is one section function curve that successively decreases.
First kind of random labelling probability curve as shown in Figure 3.Abscissa s_c and s_p represent the ratio between the burst yardstick of token quantity and token bucket, ordinate F (s_c) and F (s_p) expressive notation probability, and (S_iflx, F (S_iflx)) is the flex point of random labelling probability curve.
Relation between random labelling probability and the empty full state can also adopt second kind of relation function F ' (s) to represent:
Relation function F ' is a partitioned representation formula (s), and its variable s represents the ratio between the burst yardstick of token quantity and token bucket.When S_0≤s≤S_iflx, F ' is one section function curve that increases progressively (s).
Second kind of random labelling probability curve as shown in Figure 4.Abscissa s_c and s_p represent the ratio between the burst yardstick of token quantity and token bucket, ordinate F (s_c) and F (s_p) expressive notation probability, and (S_iflx, F (S_iflx)) is the flex point of random labelling probability curve.
Step C:, message is carried out the random drop priority flag according to described loss priority marking probability.
Produce a random number R and, relatively the size between Rand and F (s_c), the F (s_p) is carried out the loss priority mark according to comparative result to message.
If adopt first kind of random labelling probability curve, if Rand is less than F (s_p) then be red; If Rand is greater than F (s_p) but less than F (s_c), then message is yellow; If above two conditions do not satisfy, then message is green, wherein, red, yellow, green is represented high, medium and low loss priority respectively, when the system broadband is not enough, at first abandon the message that is marked as high loss priority, as red, yellow.
If adopt second kind of random labelling probability curve, if Rand is greater than F (s_p) then be red; If Rand is less than F (s_p) but greater than F (s_c), then message is yellow; If above two conditions do not satisfy, then message is green, wherein, red, yellow, green is represented high, medium and low loss priority respectively, when the system broadband is not enough, at first abandon the message that is marked as high loss priority, as red, yellow.
With first kind of random labelling probability curve is example, because the parameter of the described random labelling probability curve of influence has two classes, the control point S_0 and the S_iflx of first function, it two is coefficient A of function, therefore dynamic this two classes parameter of revising can construct dynamic marking probability curve.
Revising the curve control point method is specially: produce the another one variable V, wherein V and random parameter Rand do not possess correlation, with variable V and control point parameter S _ 0 or the S_iflx that determine by configuration for example add, subtract, simple mathematical computing such as multiplication and division, generate dynamic curve control point, expression formula according to F (s) can produce dynamic marking probability curve, and Fig. 5 is the random labelling probability curve diagram that first kind of random labelling probability curve revised the curve controlled dot generation.
Revise the function coefficients method and be specially the other one group of variable W of generation, require W and random number R and not to possess correlation equally, with variable W and the function coefficients group A that determines by configuration for example add, subtract, simple mathematical computing such as multiplication and division, produce the kinematic function coefficient sets, expression formula according to function F (s) can calculate dynamic probability curve, and Fig. 6 is that first kind of random labelling probability curve revised the random labelling probability curve diagram that function coefficients generates.
Step 204: according to the result of packet loss priority flag, token quantity in the refreshing token bucket, when needed promptly, with token quantity deduction message length corresponding token quantity in the token bucket, the new token quantity that obtains is written back in the token bucket, if quantity changes in the token bucket, then simultaneously the refresh time of token bucket is labeled as message and carries out the loss priority mark time.
If message is green, C bucket and P bucket all will be deducted message length corresponding token quantity; If message is yellow, then the P bucket will be deducted message length corresponding token quantity, and C bucket token quantity is constant; If message is red, the token quantity of C bucket and P bucket is all constant.
Realize high-precision packet wideband monitoring under the less resource prerequisite in order to be implemented in to consume, the embodiment of the invention adopts the less counter of bit wide, therefore when the time, precision was higher, counter can overflow by frequent generating period, when counter overflows, also need token bucket is refreshed, and the token bucket refresh time is recorded as counter overflows the time, specifically comprise:
Step a: the time difference between the last refresh time of the token bucket that overflows time T and record of computing counter:
C bucket refresh time is poor: Δ t
c=T-T
Cn-1
P bucket refresh time is poor: Δ t
p=T-T
Pn-1
T wherein
Cn-1Be the last refresh time of C bucket of record, T
Pn-1It is the last refresh time of P bucket of record;
Step b: according to existing token quantity K in the C bucket of record
Cn-1With existing token quantity K in the P bucket
Pn-1, and the time difference Δ t of two token buckets
cWith Δ t
pCalculate the current token quantity in C bucket and the P bucket:
C bucket token quantity: K when counter overflows
Cn=K
Cn-1+ Δ t
c* CIR
P bucket token quantity: K when counter overflows
Pn=K
Pn-1+ Δ t
p* PIR;
Step c: the token quantity of C bucket and P bucket was written back in the token bucket when counter was overflowed, and the token quantity refresh time with C bucket and P bucket is labeled as overflowing the time of counter simultaneously.
The embodiment of the invention also provides a kind of packet discarding probability tagging device, and it comprises marking probability unit and packet labeling unit.Described marking probability unit is used for the token quantity current according to token bucket, calculate the loss priority marking probability of message, described packet labeling unit is used for the packet loss priority flag probability that draws according to described probability calculation unit, and message carries out the random drop priority flag.
Described marking probability unit comprises probability function unit and probability calculation unit, and wherein said probability function unit is used to construct the random labelling probability function; Described probability calculation unit is used for according to the random labelling probability function of described probability function unit structure and the current token quantity of token bucket, calculates the loss priority marking probability of message.
Described packet labeling unit comprises random number generation unit and indexing unit, and wherein, described random number generation unit is used to produce a random number; Described indexing unit is used for the loss priority marking probability of message that described marking probability unit is drawn and random number that described random number generation unit produces compares, and message is carried out the mark of random drop priority according to comparative result.
The method and the device thereof of the packet wideband monitoring that provides by the embodiment of the invention, packet discarding probability tagging device, adopt the random labelling method that message is carried out the loss priority mark, owing to message is carried out the loss priority mark carries out at random, therefore for the message of the regular length of sharing each stream of broadband, the chance of the token that message obtains of each stream does not have fixing difference, therefore the message of a plurality of streams can be shared the broadband liberally, further, because therefore the token quantity in the equal refreshing token bucket when counter overflows can be implemented in to consume and realize high-precision wideband monitoring under the less resource.
More than the method for the packet wideband monitoring that the embodiment of the invention provided and device thereof, packet discarding probability tagging device are described in detail, used specific case herein principle of the present invention and execution mode are set forth, the explanation of above embodiment just is used for helping to understand method of the present invention and core concept thereof; Simultaneously, for one of ordinary skill in the art, according to thought of the present invention, the part that all can change in specific embodiments and applications, in sum, this description should not be construed as limitation of the present invention.
Claims (12)
1. the method for a packet wideband monitoring is characterized in that, described method comprises:
Current token quantity in the token bucket when obtaining message and carrying out priority flag;
According to token quantity current in the described token bucket, the full situation of the sky of computational token bucket expires the loss priority marking probability that situation is calculated message according to the sky of described token bucket again; The full situation of the sky of described token bucket is token quantity current in the described token bucket and the ratio between its burst yardstick; The loss priority marking probability that the full situation of described sky according to described token bucket is calculated message is to calculate by the random labelling probability function between structure random labelling probability and the empty full situation;
Loss priority marking probability and a random number of message are compared, message is carried out the loss priority mark according to comparative result;
According to the result of packet loss priority flag, the control message is to the situation that takies of bandwidth.
2. the method for packet wideband monitoring according to claim 1 is characterized in that, described when obtaining message and carrying out priority flag in the token bucket current token number measuring comprise:
Calculating is carried out loss priority mark current time and the difference of last refreshing token bucket between the time to message;
According to current token quantity in original token quantity computational token bucket in described difference and the token bucket.
3. the method for packet wideband monitoring according to claim 2, it is characterized in that, if when message is carried out the random drop priority flag, the token quantity in the token bucket changes, and then the refresh time of token bucket is updated to the time that message carries out the random drop priority flag.
4. the method for packet wideband monitoring according to claim 2 is characterized in that, if the counter of writing time overflows, then the refresh time of token bucket is updated to the time that counter overflows.
5. the method for message bandwidth supervision according to claim 1 is characterized in that, described token bucket comprises committed burst yardstick token bucket and peak value burst yardstick token bucket; The loss priority marking probability of described calculating message is specially the loss priority marking probability of the loss priority marking probability that calculates committed burst yardstick token bucket respectively and peak value burst yardstick token bucket.
6. the method for packet wideband monitoring according to claim 1, it is characterized in that, when described loss priority marking probability is when calculating gained by the decreasing function relational expression, described loss priority marking probability and random number with message compares, according to comparative result message being carried out the loss priority mark is specially: if random number is during less than the loss priority marking probability of peak value burst yardstick token bucket, message is marked as high loss priority, if but random number is during greater than the loss priority marking probability of peak value burst yardstick token bucket less than the loss priority marking probability of committed burst yardstick token bucket, message is marked as middle loss priority, otherwise message is marked as low loss priority.
7. the method for packet wideband monitoring according to claim 1, it is characterized in that, when described loss priority marking probability is when calculating gained by the increasing function relational expression, described loss priority marking probability and random number with message compares, according to comparative result message being carried out the loss priority mark is specially: if random number is during greater than the loss priority marking probability of peak value burst yardstick token bucket, message is marked as high loss priority, if but random number is during less than the loss priority marking probability of peak value burst yardstick token bucket greater than the loss priority marking probability of committed burst yardstick token bucket, message is marked as middle loss priority, otherwise message is marked as low loss priority.
8. a packet wideband monitoring device is characterized in that, described packet wideband monitoring device comprises:
Token bucket is used for deposits tokens;
Computing unit, the current token quantity of token bucket when being used for calculating message and carrying out priority flag; The marking probability unit is used for the current token quantity of token bucket that draws according to described computing unit, calculates the loss priority marking probability of message;
Described marking probability unit comprises:
The probability function unit is used to construct the random labelling probability function;
The probability calculation unit is used for the current token quantity of token bucket that random labelling probability function and described computing unit according to described probability function unit structure calculate, and calculates the loss priority marking probability of message;
The packet labeling unit is used for the packet loss priority flag probability that draws according to described marking probability unit, and message is carried out the random drop priority flag;
Described packet labeling unit comprises:
The random number generation unit is used to produce a random number;
Indexing unit is used for the loss priority marking probability of message that described marking probability unit is drawn and random number that described random number generation unit produces and compares, and according to comparative result message is carried out the loss priority mark;
The broadband control unit carries out the result of loss priority mark according to described packet labeling unit to message, and the control message takies situation to the broadband.
9. packet wideband monitoring device according to claim 8 is characterized in that, also comprises token input unit, is used for throwing in to described token bucket the token of some.
10. packet wideband monitoring device according to claim 8, it is characterized in that, also comprise the token bucket refresh unit, be used for message being carried out the result of loss priority mark and the current token quantity of token bucket that computing unit obtains, the token quantity in the refreshing token bucket according to the packet labeling unit.
11. packet wideband monitoring device according to claim 8 is characterized in that, also comprises counter, is used to write down message is carried out the time of loss priority mark and the time that token bucket refreshes; Described computing unit comprises:
Time difference calculating unit, what be used to calculate described counter records carries out time difference between the time of loss priority mark and time that token bucket refreshes to message;
The token computation unit is used for the token input speed according to token bucket, the time difference and the original token number of token bucket of described time difference computing unit calculating, the current token number when calculating is carried out the loss priority mark to message in the token bucket.
12. the device of a packet discarding probability tagging is characterized in that, described device comprises:
The marking probability unit is used for the token quantity current according to token bucket, calculates the loss priority marking probability of message;
Described marking probability unit comprises:
The probability function unit is used to construct the random labelling probability function;
The probability calculation unit is used for random labelling probability function and the current token quantity of token bucket according to described probability function unit structure, calculates the loss priority marking probability of message;
The packet labeling unit is used for the packet loss priority flag probability that draws according to described probability calculation unit, and message is carried out the random drop priority flag;
Described packet labeling unit comprises:
The random number generation unit is used to produce a random number;
Indexing unit is used for the loss priority marking probability of message that described marking probability unit is drawn and random number that described random number generation unit produces and compares, and message is carried out the mark of random drop priority according to comparative result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2007100900433A CN101043457B (en) | 2007-03-21 | 2007-03-21 | Packet wideband monitoring method and its apparatus, packet discarding probability tagging device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2007100900433A CN101043457B (en) | 2007-03-21 | 2007-03-21 | Packet wideband monitoring method and its apparatus, packet discarding probability tagging device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN101043457A CN101043457A (en) | 2007-09-26 |
| CN101043457B true CN101043457B (en) | 2010-07-07 |
Family
ID=38808662
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2007100900433A Expired - Fee Related CN101043457B (en) | 2007-03-21 | 2007-03-21 | Packet wideband monitoring method and its apparatus, packet discarding probability tagging device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101043457B (en) |
Families Citing this family (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104113487B (en) * | 2013-04-22 | 2019-04-23 | 中兴通讯股份有限公司 | The method and monitoring client of flow control |
| CN106843170B (en) * | 2016-11-30 | 2019-06-14 | 浙江中控软件技术有限公司 | Method for scheduling task based on token |
| CN112291159A (en) * | 2019-07-24 | 2021-01-29 | 中兴通讯股份有限公司 | Flow monitoring method and related equipment |
| CN112737970A (en) * | 2019-10-28 | 2021-04-30 | 华为技术有限公司 | Data transmission method and related equipment |
| CN112838959B (en) * | 2019-11-22 | 2023-03-24 | 武汉神州数码云科网络技术有限公司 | Dynamic network flow control method |
| CN112995059A (en) * | 2019-12-12 | 2021-06-18 | 阿里巴巴集团控股有限公司 | Flow control method, flow control device, terminal equipment and computer storage medium |
| CN111901248B (en) * | 2020-07-27 | 2022-05-27 | 新华三信息安全技术有限公司 | Load balancing method, device, equipment and machine readable storage medium |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1536815A (en) * | 2003-04-03 | 2004-10-13 | 华为技术有限公司 | Method for making message flow limitation by adopting token leakage cylinder |
| CN1607767A (en) * | 2003-10-16 | 2005-04-20 | 中兴通讯股份有限公司 | Flow-control method based on network processor |
| EP1249103B1 (en) * | 2000-07-18 | 2005-12-21 | Alcatel | Domain based congestion management |
-
2007
- 2007-03-21 CN CN2007100900433A patent/CN101043457B/en not_active Expired - Fee Related
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP1249103B1 (en) * | 2000-07-18 | 2005-12-21 | Alcatel | Domain based congestion management |
| CN1536815A (en) * | 2003-04-03 | 2004-10-13 | 华为技术有限公司 | Method for making message flow limitation by adopting token leakage cylinder |
| CN1607767A (en) * | 2003-10-16 | 2005-04-20 | 中兴通讯股份有限公司 | Flow-control method based on network processor |
Non-Patent Citations (4)
| Title |
|---|
| 王茜等.区分服务网络的协作可调整RED算法.电子与信息学报25 2.2003,25(2),158-163. |
| 王茜等.区分服务网络的协作可调整RED算法.电子与信息学报25 2.2003,25(2),158-163. * |
| 王茜等.基于两色标记器的协作可调整RED算法.北京邮电大学学报25 3.2002,25(3),7-11. |
| 王茜等.基于两色标记器的协作可调整RED算法.北京邮电大学学报25 3.2002,25(3),7-11. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN101043457A (en) | 2007-09-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101043457B (en) | Packet wideband monitoring method and its apparatus, packet discarding probability tagging device | |
| CN100574278C (en) | The method of refreshing token bucket and device in the flow limiting technology | |
| CN102035732B (en) | Service scheduling method and device | |
| US20140112147A1 (en) | Refresh mechanism for a token bucket | |
| CN102710514B (en) | Message flow control method and system | |
| CN107743099B (en) | Data stream processing method, device and storage medium | |
| CN103795640B (en) | A kind of flux monitoring method and device | |
| CN101599905B (en) | Method, device and system for realizing addition of traffic shaping token | |
| CN101771598B (en) | Communication dispatching method of real-time Ethernet | |
| JP2009147833A (en) | Buffer device using token bucket, and program | |
| US7333436B2 (en) | Network device with traffic shaping functions and bandwidth control method using leaky bucket algorithm | |
| CN106375238A (en) | Flow supervision method and apparatus | |
| Li et al. | Revenue maximization for communication networks with usage-based pricing | |
| CN106254272B (en) | Weighted-fair scheduling method and scheduling system based on hard-wired elasticity | |
| CN110046891A (en) | A kind of intelligence block chain interacted system | |
| CN104158538B (en) | Baud rate generation method, Baud rate generator and communication equipment | |
| CN110912834A (en) | Flow control method based on token bucket algorithm | |
| CN109905331A (en) | Array dispatching method and device, communication equipment, storage medium | |
| CN100450103C (en) | Method and device for flux plastic | |
| CN105868318A (en) | Multimedia data type prediction method and device | |
| Lu et al. | Optimal distributed scheduling of real-time traffic with hard deadlines | |
| CN105760607B (en) | The emulation component and method of emulation bus effective bandwidth based on token bucket | |
| Cheng et al. | Research on Asymmetry of Two-Queue Cycle Query Threshold Service Polling System | |
| Listova et al. | Markov's Model of the Different-Priority Traffic Routing Control in a Sensor Node | |
| CN103294903B (en) | The computer simulation method of the Poisson process of network service flow |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20100707 Termination date: 20170321 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |