CN1592267A - Priority message flow ensuring method when network disaster - Google Patents
Priority message flow ensuring method when network disaster Download PDFInfo
- Publication number
- CN1592267A CN1592267A CNA031563244A CN03156324A CN1592267A CN 1592267 A CN1592267 A CN 1592267A CN A031563244 A CNA031563244 A CN A031563244A CN 03156324 A CN03156324 A CN 03156324A CN 1592267 A CN1592267 A CN 1592267A
- Authority
- CN
- China
- Prior art keywords
- disaster
- priority
- message
- bandwidth
- cover
- 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.)
- Granted
Links
Images
Landscapes
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
This invention relates to a method for guaranteeing priority message flow at network catastrophe including: before the messages enter into the array, the catastrophic state is judged first, a set of bandwidth parameters matched for a set of priority array of interface outlet port at normal state in used, the messages enter into the set array according to the degree of the priority. At the catastrophe states a set of catastrophic bandwidth parameters for catastrophic state is used and messages enter into the said array in terms of the priority degree.
Description
Technical field
The present invention relates to a kind of network technology, in the network of forming by communication equipment, when catastrophe failure-when being referred to as network meltdown taking place, the assurance technology of network medium priority message flow, i.e. a kind of reservation techniques of the network bandwidth.
Background technology
In the network that connects to form by communication equipment, when fault such as opening circuit, because of rerouting can cause the variation of flow in the network, take place unavoidably congested, when serious even cause the priority packet packet loss.
The packet loss problem of this priority packet is described below in conjunction with RPR network (RPR:RESILIENT PACKET RING).
The Resilient Packet Ring network technology is a kind of emerging looped network technology that has self-healing function since it have efficiently, safety, advantage such as stablize, thereby obtained extensive use.
The RPR ring is the radio network of similar Ethernet, compares with Ethernet token ring (Token Ring) in the past, and the efficient of RPR ring is than higher.Being mainly reflected in token ring (Token Ring) is the source deletion, and the RPR ring is the purpose deletion, promptly when a unicast message transmits on ring, the purpose deletion is that message arrival destination is later on just deleted from encircling, for example the message of AB will be deleted at the B place among Fig. 1, can not take the bandwidth of BCDA; After source deletion then needs message is sent to the source node that sends message along ring, just deleted, such as the message of AB among Fig. 1, need process through ABCDA, could be deleted.
The interior ring and the outer shroud of RPR ring all are unidirectional traffics, transmit data to fixing direction.As interior ring 11 and outer shroud 12 among Fig. 1, the unidirectional traffic in the interior ring 11 is anticlockwise, and the unidirectional traffic in the outer shroud 12 is clockwise.
The another one function of RPR ring is the self-healing function when breaking down, promptly when the some nodes on the RPR ring break down, flow as the original AC of supposition among Fig. 2 is walked outer shroud 12 (ADC), when the optical fiber between D, the C opens circuit, the D node can be automatically with flow loopback (Wrap), ring 11 in will going into from the flow loopback of outer shroud 12 promptly temporarily is sent to C node (DABC) by interior ring 11 with data message.
After the loopback (Wrap), node device can select ring (ring or outer shroud in walking) again according to the network topological information of newly learning, and behind the choosing ring, ring 11 transmitted from ABC in the flow of AC will be modified to away, shown in arrow among Fig. 2 13 again; Former cause D node sends to the flow of C node, also becomes through interior ring 11 to send to C from DABC, shown in arrow among Fig. 2 14.
In above troubleshooting, flow becomes the ABC process from ADC, can superpose with original flow from B to C, if the flow-rate ratio after the stack is bigger, will produce congested at the B node.Because general RPR ring all is core net or backbone network, also is that the utilance of network is higher usually, above-mentioned fault treating procedure can cause the B point to produce very serious congestion situation, under the extreme case, may be 2 times of RPR monocycle flow even more.In loopback is switched, because some packet losses can take place in the variation of flow inevitably, because after choosing ring again (flow walk in encircle after), network may all keep moving in this state within for a long time, when waiting connection of breaking or the node that breaks down to be resumed always.How under so congested situation, guarantee various messages particularly the high priority message be not dropped, need take seriously.
Present ensuring method is undertaken by algorithm, drops into the fair algorithm technology of using that Cisco is arranged.This fair algorithm, when having fault, also do not come into force (when not having fault, general congested also not serious).Also illustrate with top example.When the RPR ring did not have fault, if A → D → C has flow (being referred to as Fac), D → C also had flow (being referred to as Fdc) simultaneously.When two flow sums go up the total bandwidth of transmission greater than ring, will produce congested at node D.Fair algorithm will temporarily suppress the flow that node D goes up ring, simultaneously send congested back-pressure message to node A, tell A that node link is congested, the A node will reduce the Fac flow after receiving this congested back-pressure message, thereby solve the congestion problems on the D node, what guarantee to encircle is unimpeded.
Handle in the above congested, suppress will to suppress low-priority traffic in the flow of node D, do not suppress high-priority traffic, guarantee that with this high priority message can not be dropped.
In the last example, when causing the B node congested, start fair algorithm equally when breaking down on the ring, with the flow that temporarily suppresses to encircle on the Node B, by send congested back-pressure message to node A, reduce the Fac flow simultaneously, thereby solve the congestion problems on the B node, what guarantee to encircle is unimpeded.
Therefore fair algorithm uses the good algorithm of can yet be regarded as under normal discharge.
But in actual applications, when the initial ring disaster scenario, as D, optical fiber between the C opens circuit, and Fac (sending to the flow of C from A through B) and Fbc (sending to the flow of C from B) are very big (such as on the ring of 2.5G, the A node has multicast message+600M high priority message+2.4G low priority message of 1G to the flow of C node, the B node is to the 400M high priority message and the 1.6G low priority message in addition of C node, according to algorithm, should abandon low priority message 2.4G+1.6G, guarantee the multicast message of high priority message 600M+400M and 1G), when also having a large amount of multicast messages (adopting the broadcast transmission form) flow simultaneously on the ring, between the C node heavy congestion appears at the B node.
At this moment use fair algorithm, the situation of high priority message and a small amount of packet loss of high priority multicast message can appear in discovery, by analysis, be owing to encircle (as 2.5G) for RPR at a high speed, the buffering area (Buff) that comprises the chip of fair algorithm on the node is big inadequately, a large amount of congested the causing that can not absorb burst.
Because the high priority message generally is important message (such as voice flow), the high priority multicast message generally is a video flowing.A small amount of packet loss may cause that video/audio is unintelligible, is not allow to take place high priority message or high priority multicast message packet loss in actual applications therefore.
We are under special congestion situation, can't guarantee in actual the realization high priority message fully not the defective of packet loss be called the A defective.Certainly this defective can solve (but needing to revise the RPR chip design) by the Buff that increases RPR chip on the node.
In addition, when the flow after the various high priority message addition on the RPR ring greater than the disaster state under during utilizable bandwidth, the situation that high priority message packet loss will inevitably occur, and for all each high priority message flows that participates in stack, all packet loss may appear.
For example among Fig. 3, suppose that this ring is the RPR ring of 2.5G.Suppose to comprise among the Fdc (D is to the flow of C) the high priority message of 1G, the high priority message that comprises 1G among the Fac, the high priority message that comprises 1G among the Fbc again, these flows are under the situation of the no disaster of ring, Fdc walks outer shroud and sends to C, C was given in environment-development in Fbc walked, no matter Fac walks interior ring through the B node, or Fac walks outer shroud through the D node, the high priority message all is no more than 2.5G, so can not be dropped.
Under the disaster state that generation optical fiber opens circuit between D, the C, D, A, B give the flow of C and walk interior ring, flow between B, the C will be greater than 2.5G (with being 3G) like this, and at this moment fair algorithm will cause 3 flows all may packet loss, and this is that the user is reluctant the result that receives.In fact, the user wishes in this state, realizes not packet loss of whole net height priority packet by reducing some high priority message flows, rather than all may packet loss.
The capacity that we encircle this high-priority traffic down above disaster scenario, the defective that packet loss all may appear in all high priority messages call the B defective.
What the present invention need solve is exactly two kinds of defectives that run into when adopting fair algorithm.Before the explanation technical solution of the present invention, also to introduce following pre-knowledge earlier.
On the interface that encircles on each node, each INTERNET service provider (ISP) has used differential service (DiffServ is to call DiffServ in the following text) to distinguish high Low Priority Queuing at present.DiffServ uses in the IP head 6 (being called the DSCP territory) in 8 TOS territories, thereby stream can be divided into 64 classes, standard has defined 6 classes: EF (Expedited Forwarding), AF1~AF4 (Assured Forwarding) and BE (Best Effort) at present, and other temporarily keeps.Wherein the EF class is a limit priority, and AF1~AF4 class is the priority of user oneself definition, and the BE class is a lowest priority.
In order to guarantee the flow of each priority packet, 6 formations have all been defined, corresponding above-mentioned respectively 6 classes stream in the interface outlet of each node of RPR.Wherein there is strict height priority relationship between EF formation and other formation, could sends the stream (adopt a kind of priority query algorithm, its technical term is the PQ algorithm, PQ:Priority Queue) of other formation after having only the EF formation to send to finish.The user disposes certain bandwidth or the weights of scheduling (Weight value) in order to guarantee priority respectively to these six kinds of formations.By using bandwidth guaranteeing property restriction or the distribution of dispatching algorithm to the different priorities formation.
Different vendor uses different dispatching algorithms, comprise that low time delay bandwidth guarantees that algorithm (is called for short the LLS algorithm, LLS:Low-latency sustainable bandwidth), general time delay bandwidth guarantees that algorithm (is called for short the NLS algorithm, NLS:Normal-latency sustainable bandwidth), the Weighted Fair Queuing algorithm (is called for short the WFQ algorithm, WFQ:weighted fair queuing) and peak rate bandwidth constraints algorithm (be called for short PBS algorithm, PBS:Peak bandwidth service).LLS and NLS are the algorithms that the bandwidth to formation guarantees, different is: the bandwidth that the LLS algorithm guarantees has better time delay.The WFQ algorithm can distribute bandwidth according to the weights of configuration.The PBS algorithm is the algorithm of restriction maximum bandwidth, and the flow that promptly surpasses configured bandwidth all will be dropped.
Summary of the invention
Priority packet flow ensuring method when the objective of the invention is to design a kind of network meltdown, by a kind of technological means is provided to the user, solve the network that encircles a class as RPR, when disaster occurring, no matter whether the high priority message surpasses the network rings capacity, can guarantee that all the high priority message is not dropped, realize that the high-priority traffic of whole net guarantees.
The present invention is by a kind of simple and effective " disaster RSVP technology ", promptly pass through the reservation schemes of planned network bandwidth, use pre-configured bandwidth and algorithm to limit to various priority class traffics, when making RPR one class network generation disaster, guarantee that the high priority message can not be dropped, to guarantee the flow of whole net priority packet.
The technical scheme that realizes the object of the invention is such: the priority packet flow ensuring method during a kind of network meltdown is characterized in that comprising following treatment step:
A. on the node device interface outbound port of network, before message is gone into formation, judge current disaster generation state earlier;
B. under the normal condition that no disaster takes place, use a cover bandwidth parameter that overlaps priority query's configuration under the normal condition as one of interface outbound port, under the normal bandwidth parameter of configuration, message is just gone into this cover formation by flow priority;
C. under the disaster state that has disaster to take place, use a cover disaster bandwidth parameter that overlaps priority query's configuration under the disaster state as one of interface outbound port, under the disaster bandwidth parameter of configuration, message is just gone into this cover formation by flow priority.
Can have only a socket outbound port priority query, under normal condition, use the normal bandwidth parameter of configuration, under the disaster state, use the disaster bandwidth parameter of configuration; Two socket outbound port priority queries also can be set, and another set of formation of disposing the disaster bandwidth parameter is used in the formation of using a cover to dispose the normal bandwidth parameter under normal condition under the disaster state.
Under the disaster state, also can on the node device interface arrival end of network, use and/or increase and/or change message flow rule and, be used to change and adjust the path of priority of messages, bandwidth, forwarding the action of message flow.
The present invention carries out on the node interface of network meltdown RSVP at needs, assurance bandwidth and maximum (peak value) bandwidth by high-priority queue in the formation of configuration disaster, limit individual nodes under disaster scenario on the high priority bandwidth of ring, thereby realize disaster scenario down ring go up the high priority bandwidth and can not surpass band width in physical.
But the inventive method provides a kind of means to the user, when having solved the network that encircles a class at RPR and disaster occurring, is no more than the ring capacity when the high priority message under the congested especially situation, abandons the low priority message as far as possible, guarantees that the high priority message is not dropped; When the high priority message surpasses the ring capacity, allow certain node in the flow to be limited, thereby the high priority of realizing whole net guarantees by the pre-configured bandwidth parameter of user.
The beneficial effect of the inventive method is to have solved at RPR to encircle under the situation of a class network generation disaster, because two defectives-A defective and the B defective of using fair algorithm to bring.
Can not guarantee down the high priority message defective A of packet loss not fully for special congestion situation, the inventive method can be by disposing peak value service speed algorithm (PSR:PeakService Rate) to disaster formation low priority message, maximum bandwidth by the restriction low-priority traffic solves, the user can be according to total possible bandwidth (B1) of high priority message and the band width in physical (B2) on the network rings, draw the message bandwidth (B3) that allows to go up the low priority that encircles, generally can adopt B3 to carry out less times greater than the method for (B2-B1).Because fair algorithm still works to a certain extent, so total possible bandwidth (B1) of permission high priority message can be greater than band width in physical (B2), for the equipment of different model, the ratio of the shared bandwidth of each several part message need decide according to the empirical value of actual tests.LLS/NLS and PBS, WFQ algorithm can use together, can realize both having guaranteed certain bandwidth (using LLS or NLS), limit its maximum bandwidth (using PBS simultaneously) again.Relevant algorithm can be with reference to the relevant document of introducing, because its using method has exceeded the technical scope that the present invention relates to, so no longer describe in detail.
Under the disaster state, the defective B of packet loss all may appear in all high priority messages when high-priority traffic surpassed the ring capacity, the inventive method, can be by the assurance bandwidth and maximum (peak value) bandwidth of high priority in the formation of configuration disaster, limit individual nodes under disaster scenario on the ring the high priority bandwidth, thereby the realization disaster scenario encircles upward down, and the high priority bandwidth can not surpass band width in physical, guarantees the transmission of most important business.
Technology of the present invention has been to increase the operation after the discovery network failure, and before the present invention, generic failure only adopts alarm when taking place, and the inventive method is then gone back initiatively employing and node carried out operations such as bandwidth constraints.
Description of drawings
Fig. 1 is a RPR ring structure schematic diagram;
Fig. 2 is under the RPR ring failure condition, the flow situation schematic diagram of node A to C and node D to C;
Fig. 3 is under the RPR ring failure condition, Fdc, Fac, Fbc flow situation schematic diagram;
Fig. 4 is the formation switching flow schematic diagram under the disaster state.
Embodiment
The basic principle of the inventive method foundation is: when disaster (fault) takes place ring (network), use pre-configured bandwidth and strategy to limit to low-priority traffic, thereby guarantee the transmission of high-priority traffic.
When ring breaks down, the fault on the very fast discovery ring of the consultation of RPR self, and know loopback (Wrap) taken place that beginning is the choosing ring again.After choosing encircles again, enter the stable state under a kind of malfunction, at this moment ring is gone up the bandwidth before actual utilizable bandwidth has been far smaller than fault, and generally speaking, available bandwidth will be smaller or equal to 1/2 of former bandwidth.But at this moment, the outbound port formation of RPR node interface of ring on each actually still comes scheduling parameter such as configured bandwidth assurance according to the configuration before the fault, and this also is to cause ring to go up congested serious basic reason.
In the inventive method, on each need carry out the node interface of network meltdown RSVP, 12 formations are set.Wherein 6 is the normal formation of using (EF, AF1-AF4 and BE) of user, is referred to as normal queue.Other 6 formations for using under the disaster state are referred to as the disaster formation.The user can dispose the bandwidth parameter of oneself in the disaster formation.These bandwidth parameters can just be configured at the beginning of network is set up according to the situation of planning.
During RPR network generation disaster, agreement by RPR self can be learnt this disaster very soon, be handed down to service plane (Forwarding plane) then, this Forwarding plane will change the formation that message is gone into, be about to message and go into corresponding disaster formation, like this, the parameter that is disposed for the disaster formation will play a role, to guarantee that service bandwidth can transmit according to user's demand under the disaster state.
Service plane (Forwarding plane) identifies as follows to going into the modification of formation with pseudo-code:
If (the disaster generation==True)
Message is gone into corresponding disaster DiffServ formation
Else
Message is gone into normal DiffServ formation
Being revised as of control plane:
(1) if RPR network generation disaster then is provided with the disaster generation flag bit of service plane;
(2) distribution mechanism of increase disaster formation and permission configured bandwidth mechanism.
Above-mentioned Forwarding plane is the plane that data service is handled, and control plane is by assigning the operation of commands for controlling Forwarding plane.
How the following describes Forwarding plane changes message and goes into formation.A lot of formations are arranged in each network processing unit, can pass through configuration register, formation is assigned to certain port.It is exactly that the message of this formation will send from this port that formation is assigned to certain port.Existing each interface has 6 fixing formations, and 6 formations are that order increases, and each interface has the plot of a formation.Such as interface 1, suppose first formation since 129, plot is exactly 128 so, first formation=plot+port queue skew.
When using method of the present invention, can be 12 formations of each interface assignment, the parameter of preceding 6 formations configuration is the parameter (such as guaranteeing bandwidth, maximum bandwidth or the like) under the user configured normal condition, other 6 formations are no at ordinary times, be the disaster formation under the disaster scenario, the user can pre-configured disaster scenario under the parameter (such as guaranteeing bandwidth, maximum bandwidth or the like) of these formations.
In conjunction with referring to Fig. 4, before message is gone into formation, need to judge disaster generation flag bit (step 41) that when this flag bit is 0, show that disaster does not take place, message will be gone into normal formation (step 43).When disaster takes place,, can learn disaster very soon and take place, then with this mark position 1 by the agreement of RPR oneself.When the judgement symbol position was 1, message was gone into the disaster formation, thereby had finished the switching (step 42) of formation, so just can use the parameter of configuration under the disaster scenario.
Said process is as follows with the pseudo-code sign:
Control plane:
If (disaster generation)
{ Forwarding plane disaster flag=1 is set; }
else
{ disaster Flag=0; }
If (disaster answer)
{ Forwarding plane disaster flag=0 is set; }
Datum plane:
If (disaster flag==0)
{ message is gone into normal queue; }
Else
Message go into formation number=message go into formation number+6;
Message is gone into formation (disaster formation)
}
Control plane is revised its (1) embodiment in said process, is exactly after finding disaster, and the respective flag position of datum plane is set.
The inventive method, the disaster (fault) of the network of forming to the important traffic equipment that can predict or by communication equipment is provided with the parameter that a cover is fit to disaster (fault), and use when disaster takes place reaches and wishes the result that reaches under the disaster scenario; For this particular network of RPR, by the bandwidth parameter under disaster (fault) situation is set, guarantee under disaster (fault) situation, the transmission of important service, perhaps the certain user wishes the specific purposes that reach; Under communication equipment or network meltdown situation, same priority is in other words with a kind of message of business, when it surpasses actual transfer capability under the disaster scenario, can pass through preset parameters or allocation plan, again bandwidth, priority etc. is configured, reaches and guarantee the more purpose of important service; Parameter under the disaster scenario that is disposed is not under normal circumstances used, and only uses when disaster takes place.
Illustrate RSVP technology of the present invention below with reference to Fig. 3.
Under the normal condition, A → C and D → C walk outer shroud, and B → C walks interior ring, suppose and have only B → C and A → C at present, three flows of D → C.The EF formation is the highest priority data formation, is generally used for transmitting business such as voice, image.For this routine networking, the maximum bandwidth of a ring permission is 2.5G.
Under the normal condition, the parameter of formation is as shown in table 1 on the configuration A node R PR interface:
Formation | Guarantee bandwidth | Maximum bandwidth |
????EF | ?????1000M | ?????1000M |
????AF1 | ?????100M | |
????AF2 | ?????100M |
Table 1
Under the normal condition, the parameter of formation is as shown in table 2 on the configuration D node R PR interface:
Formation | Guarantee bandwidth | Maximum bandwidth |
????EF | ?????1000M | ?????1000M |
????AF1 | ?????100M | |
????AF2 | ?????100M |
Table 2
Under the normal condition, the parameter of formation is as shown in table 3 on the configuration B node R PR interface:
Formation | Guarantee bandwidth | Maximum bandwidth |
?EF | ??1000M | ??1000M |
?AF1 | ??500M | ??700M |
?AF2 | ??500M | ??800M |
Table 3
Under the normal condition, when A → C and D → C walked outer shroud, the bandwidth traffic that guarantees in the ring was 1000M+100M+100M+1000M+100M+100M=2400M<2.5G, can transmit.When encircling in B → C walks, the assurance bandwidth traffic in the ring is: 1000+500+500=2000M<2.5G, also can transmit.
Under the disaster scenario, the parameter of formation is as shown in table 4 on the configuration A node R PR interface:
Formation | Guarantee bandwidth | Maximum bandwidth |
?EF | ?500M | ?500M |
?AF1 | ?0 | |
?AF2 | ?0 |
Table 4
Under the disaster scenario, the parameter of formation is as shown in table 5 on the configuration D node R PR interface:
Formation | Guarantee bandwidth | Maximum bandwidth |
?EF | ?1000M | ?1000M |
?AF1 | ?0 | |
?AF2 | ?0 |
Table 5
Under the disaster scenario, the parameter of formation is as shown in table 6 on the configuration B node R PR interface:
Formation | Guarantee bandwidth | Maximum bandwidth |
?EF | ?1000M | ?1000M |
?AF1 | ?0 | |
?AF2 | ?0 |
Table 6
If do not use method of the present invention, D is taking place, under the malfunction that opens circuit between the C, above flow A → C, D → C, and B → C will walk interior ring, so total EF flow=1000+1000+1000=3000M>2.5G.Though adopt fair algorithm that AF stream is dropped, when EF flows greater than the capacity that encircles, with abandoning of justice from A, B, the EF flow of D node, the flow of such three nodes all can have packet loss.
After adopting method of the present invention, if think A node inessential (referring to circulation), can limit A, tolerance abandons the EF flow (as shown in table 4) on a part of A, this moment is EF flow=500+1000+1000=2500M=2.5G always, make total EF stream still within endless belt is wide like this, to guarantee B, the D flow is normal.
In the above-described embodiment, be to use the different formation of two covers to reach RSVP purpose of the present invention, in actual the realization, it is also conceivable that other scheme of employing, comprising:
Can only use a cover formation, but be to use two covers (or more than two covers) parameter.Use a cover normal bandwidth parameter at ordinary times under the normal condition, when network meltdown takes place, use pre-configured disaster bandwidth and dispatching algorithm parameter, also can reach effect same.But software design is slightly complicated.
Its specific implementation is: the same under queue assignment and the normal condition in advance for formation configures queue parameter under the disaster scenario, or each interface is established 6 formations.But at control plane, for each port keeps two cover parameters, a cover is the parameter under the user configured normal condition, and another set of is the user configured parameter that is used under the disaster scenario.When disaster takes place, learn that by the RPR agreement disaster takes place equally, control plane will be configured to each formation with the queue parameter under the disaster scenario (guaranteeing bandwidth, peak bandwidth etc.) then.The parameter of each formation use has been exactly the parameter under the disaster scenario like this.After replying normally, the parameter under the configuration normal condition has so just been replied the parameter configuration under the normal condition again automatically.
Be expressed as follows with the pseudo-code sign:
Control plane:
If (disaster generation)
{ issuing the disaster bandwidth parameter } to datum plane
If (disaster answer)
{ issuing the normal bandwidth parameter } to datum plane
Datum plane: need not revise
Control plane is revised its (2) embodiment in said process, is exactly after finding disaster, the queue parameter on configuration data plane.
The inventive method can also be expanded when implementing, define as rank disaster (fault), different disaster (fault) rank is used different parameters or different disaster allocation plans, perhaps use different formation (if formation is abundant).For example: adopt allocation plan 1 when RPR encircles a node failure; Adopt allocation plan 2 or the like during two node failures.For communication equipment or network, the situation of multiple rank disaster (fault) is arranged, many cover parameters or scheme can be set; According to the actual rank that takes place of disaster (fault), use corresponding parameters or scheme automatically.For the perception of fault rank and malfunction, can also can directly be provided with by communication equipment by automatic perception of means such as agreement, signals by artificial.
Further, when disaster takes place, also can be as required, on the node device interface arrival end of network, use and/or increase and/or change for normal condition part or all of message flow rule and to the action of message flow, be used to change and adjust the path of priority of messages, bandwidth even forwarding.The feature that these message flow rules are had by message itself and transmit in feature various key elements and their combination in any (comprising mask mode and scope mode etc.) formed for the differentiation data flow, comprise the source interface of message, the purpose interface, source physics (MAC) address, the target physical of message (MAC) address, IP message source address, IP message destination address, network protocol type, transmission control protocol (TCP) source port, transmission control protocol (TCP) destination interface, User Datagram Protoco (UDP) (UDP) source port, User Datagram Protoco (UDP) (UDP) destination interface, whether be fragment message and whether be transmission control protocol (TCP) sync message etc.Action to message flow, comprise filtration (abandoning) message, to message set up new registration procedure priority (Remark), to message conduct interviews rate limit (CAR:Committed Access Rate), implementation strategy route and be redirected etc. and can be used for all actions of communication equipment data flow and their combination.
For example when disposing under the disaster scenario on the A node R PR interface parameter of formation for example, also can carry out finer division to the flow on the A above-mentioned.From a plurality of ports, the port that has is to connect important client such as the flow that encircles on the A node, and the port that has is to connect not too important customers.When the EF formation limits on to A, can reduce the EF priority of messages by the source port of identification flow, reducing its priority such as the EF stream with inessential port is AF1 stream, like this, on the A node, also can realize assurance to the high priority flows of Very Important Person.
Come by the popularization of RPR looped network,, also can adopt method of the present invention for the network failure of other communication apparatus node and composition thereof.
The inventive method also can be the user more selection is provided when implementing, make the user realize controlled to the bandwidth under the network meltdown state.For the transmission that guarantees high-priority service on the network, the reliability that improves network is significant.
The inventive method can also be generalized in the solution of other similar problem, adopt to distinguish the method for malfunction and normal condition, in advance disaster (fault) state that may occur is prevented and planning in advance, thereby initiatively, effectively deal with problems.
During enforcement, can also limit that means such as (CAR:Committed Access Rate) or other bandwidth guarantee, the bandwidth constraints algorithm is realized the bandwidth constraints among the present invention, not necessarily above-described LLS, NLS scheduling algorithm by access rate.
Claims (12)
1. the priority packet flow ensuring method a during network meltdown is characterized in that comprising following treatment step:
A. on the node device interface outbound port of network, before message is gone into formation, judge current disaster generation state earlier;
B. under the normal condition that no disaster takes place, use a cover bandwidth parameter that overlaps priority query's configuration under the normal condition as one of interface outbound port, under the normal bandwidth parameter of configuration, message is just gone into this cover formation by flow priority;
C. under the disaster state that has disaster to take place, use a cover disaster bandwidth parameter that overlaps priority query's configuration under the disaster state as one of interface outbound port, under the disaster bandwidth parameter of configuration, message is just gone into this cover formation by flow priority.
2. the priority packet flow ensuring method during network meltdown according to claim 1, it is characterized in that: in the described steps A, judge that current disaster generation state is a state of judging disaster generation flag bit, when learning the disaster generation, procotol, when procotol is learnt the disaster answer, disaster generation flag bit is resetted to the set of disaster generation flag bit.
3. the priority packet flow ensuring method during network meltdown according to claim 1 is characterized in that: one of interface outbound port cover priority query is a cover formation of using under the normal condition among the described step B; One of interface outbound port cover priority query among the described step C is a cover disaster priority query that sets up, and this cover disaster priority query uses a cover disaster bandwidth parameter that is disposed.
4. the priority packet flow ensuring method during network meltdown according to claim 1 is characterized in that: in the described steps A, also comprise and judge current disaster generation rank; One of interface outbound port cover priority query is the formation of using under the normal condition among the described step B; One of interface outbound port cover priority query among the described step C, it is a cover formation of from the many covers disaster priority query that sets up, selecting by disaster generation rank, every suit disaster priority query configuration one cover disaster bandwidth parameter, a kind of disaster generation rank is corresponding with a cover disaster priority query.
5. the priority packet flow ensuring method during network meltdown according to claim 1, it is characterized in that: among described step B and the described step C, one cover priority query of interface outbound port is same set of formation, under the normal bandwidth parameter of using configuration under the normal condition of step B and disaster state, use the disaster bandwidth parameter of configuration at step C.
6. the priority packet flow ensuring method during network meltdown according to claim 5 is characterized in that: described disaster bandwidth parameter has more than the cover, and is corresponding with the rank of the actual generation of disaster respectively.
7. the priority packet flow ensuring method during network meltdown according to claim 1 is characterized in that: described bandwidth disaster parameter comprises the dispatching algorithm that assurance bandwidth under the various priority queries and maximum bandwidth and bandwidth guarantee.
8. the priority packet flow ensuring method during network meltdown according to claim 1, it is characterized in that: described bandwidth disaster parameter is pre-configured on the important traffic node device of network, perhaps needs to carry out on the node device of network meltdown RSVP.
9. the priority packet flow ensuring method during network meltdown according to claim 1, it is characterized in that: further comprise among the described step C, cover priority query to described interface outbound port is reduced to message on the Low Priority Queuing with the message on its highest-priority queue.
10. the priority packet flow ensuring method during network meltdown according to claim 1, it is characterized in that: also comprise among the described step C, on the node device interface arrival end of network, use and/or increase and/or change message flow rule and, be used to change and adjust the path of priority of messages, bandwidth, forwarding the action of message flow.
11. the priority packet flow ensuring method during network meltdown according to claim 10 is characterized in that: the feature that described message flow rule is had by message itself and transmit in feature forms for the various key elements of differentiation data flow and their combination in any.
12. the priority packet flow ensuring method during network meltdown according to claim 10, it is characterized in that: described action message flow, comprise filtering packets, to message set up new registration procedure priority, to message conduct interviews rate limit, implementation strategy route and be redirected.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CNB031563244A CN100414916C (en) | 2003-09-03 | 2003-09-03 | Priority message flow ensuring method when network disaster |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CNB031563244A CN100414916C (en) | 2003-09-03 | 2003-09-03 | Priority message flow ensuring method when network disaster |
Publications (2)
Publication Number | Publication Date |
---|---|
CN1592267A true CN1592267A (en) | 2005-03-09 |
CN100414916C CN100414916C (en) | 2008-08-27 |
Family
ID=34598379
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CNB031563244A Expired - Fee Related CN100414916C (en) | 2003-09-03 | 2003-09-03 | Priority message flow ensuring method when network disaster |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN100414916C (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101682553A (en) * | 2008-03-31 | 2010-03-24 | 松下电器产业株式会社 | Communication terminal device and communication control method |
CN101390336B (en) * | 2005-03-10 | 2011-11-02 | 意大利电信股份公司 | Disaster recovery architecture |
CN102368736A (en) * | 2011-11-10 | 2012-03-07 | 华为技术有限公司 | Message sending method and equipment |
CN112087257A (en) * | 2019-06-13 | 2020-12-15 | 华为技术有限公司 | Fault protection method, equipment and system of optical network |
CN113906720A (en) * | 2019-06-12 | 2022-01-07 | 华为技术有限公司 | Traffic scheduling method, device and storage medium |
Family Cites Families (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6594232B1 (en) * | 1999-06-02 | 2003-07-15 | Marconi Communications, Inc. | Transmitter-based path protection switching in a ring network |
WO2002017552A1 (en) * | 2000-08-24 | 2002-02-28 | Ocular Networks | Apparatus and method for facilitating data packet transportation |
ITMI20012088A1 (en) * | 2001-10-10 | 2003-04-10 | Cit Alcatel | METHOD FOR PROPAGING FAULT INFORMATION IN A RPR NETWORK AND RELATED TYPE OF RPR PACKAGE |
JP3705222B2 (en) * | 2002-02-06 | 2005-10-12 | 日本電気株式会社 | Path setting method, communication network using the same, and node device |
-
2003
- 2003-09-03 CN CNB031563244A patent/CN100414916C/en not_active Expired - Fee Related
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101390336B (en) * | 2005-03-10 | 2011-11-02 | 意大利电信股份公司 | Disaster recovery architecture |
CN101682553A (en) * | 2008-03-31 | 2010-03-24 | 松下电器产业株式会社 | Communication terminal device and communication control method |
CN102368736A (en) * | 2011-11-10 | 2012-03-07 | 华为技术有限公司 | Message sending method and equipment |
CN102368736B (en) * | 2011-11-10 | 2014-12-10 | 华为技术有限公司 | Message sending method and equipment |
CN113906720A (en) * | 2019-06-12 | 2022-01-07 | 华为技术有限公司 | Traffic scheduling method, device and storage medium |
CN113906720B (en) * | 2019-06-12 | 2024-05-10 | 华为技术有限公司 | Traffic scheduling method, traffic scheduling device and storage medium |
CN112087257A (en) * | 2019-06-13 | 2020-12-15 | 华为技术有限公司 | Fault protection method, equipment and system of optical network |
WO2020248634A1 (en) * | 2019-06-13 | 2020-12-17 | 华为技术有限公司 | Fault protection method, device and system for optical network |
CN112087257B (en) * | 2019-06-13 | 2022-04-12 | 华为技术有限公司 | Fault protection method, equipment and system of optical network |
Also Published As
Publication number | Publication date |
---|---|
CN100414916C (en) | 2008-08-27 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN1294728C (en) | Method and system for providing QoS assurance in edge router | |
US7327675B1 (en) | Fairness of capacity allocation for an MPLS-based VPN | |
US6996062B1 (en) | Policy-based weighted random early detection method for avoiding congestion in internet traffic | |
US6748435B1 (en) | Random early demotion and promotion marker | |
CN1606850A (en) | Bandwidth reservation reuse in dynamically allocated ring protection and restoration technique | |
CN1863144A (en) | Method for providing differential service | |
CN101106518B (en) | Service denial method for providing load protection of central processor | |
US8547846B1 (en) | Method and apparatus providing precedence drop quality of service (PDQoS) with class-based latency differentiation | |
CN1701319A (en) | Edge-based per-flow QoS admission control in a data network | |
CN1774708A (en) | Methods and devices for regulating traffic on a network | |
CN1649320A (en) | System and its method for guaranteeing service quality in virtual special net based network | |
CN1420665A (en) | Packet sending device and packet sending processing method | |
CN1449162A (en) | Telecommunications system employing virtual service network architecture | |
JP2004260832A (en) | Method for providing service with guaranteed quality of service in ip access network | |
CN1820462A (en) | Systems and methods for alleviating client over-subscription in ring networks | |
CN1825798A (en) | Method for implementing information stream transmission in next generation network | |
US8203956B1 (en) | Method and apparatus providing a precedence drop quality of service (PDQoS) | |
CN1744563A (en) | Method for realizing strate gic route in Ethernet switch | |
WO2012161868A1 (en) | Non-Uniform Per-Packet Priority Marker For Use With Adaptive Protocols | |
CN101064637A (en) | Method for guaranteeing quality of service of operation maintenance data stream | |
CN1773952A (en) | Bandwidth gurantee method for communication equipment or high-priority messaging during network at fault | |
CN1852086A (en) | Binding interface in RPR and its realizing method and apparatus | |
CN1798098A (en) | Method of differentiating multiple services for anti IP data stream in multicast to impact communication system | |
CN100338911C (en) | Method and apparatus for dynamic bandwidth allocation over an internet protocol telecommunications network | |
CN1592267A (en) | Priority message flow ensuring method when network disaster |
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: 20080827 Termination date: 20150903 |
|
EXPY | Termination of patent right or utility model |