DE19634271A1 - Digital network with time dependent resource planning - Google Patents

Abstract

The digital network includes multiple nodes (1-9) with physical lines for handling resource transactions (R1-R8) between customers and suppliers. The nodes are coupled to a controller (10) which runs network management software. Resource data is stored in a memory of a resource planning unit and is accessed by the controller. Preferably, the resource planning unit has entries at preset times for the applications of the data.

Description

The invention relates to a network with several nodes, with several the lines connecting the nodes and with a control circuit.

Such a network, which for example an SDH network (SDH = Syn Chrone digital hierarchy) contains several nodes, each with ver different physical lines are connected. The lines can each have different transmission capacities and a signal or transport a multiplex signal. A connection is used as the logical path between two nodes or endpoints with a certain transmission capacity and quality of service. A path can have several Intermediate nodes run. Likewise, between a physical line two nodes run several logical paths. To determine the knots for a path becomes a path search, which is also called a route, by means of a Control circuit, which contains a network management software performed. The Control circuitry also receives orders from customers who have a particular conn Example, between two nodes in the network at a certain time score and wish with a certain transmission capacity.

The invention is therefore based on the object of creating a network that Receive and process orders from customers.

The task is for the above network, which has multiple nodes, contains several lines connecting the nodes and a control circuit solved the following measures:  

The control circuit is

• - to determine a path with the resources contained in nodes for an order is provided which is at least determined by the end points, the start time, the duration and transmission capacity of a connection is
• - for the time-dependent storage of the resources assigned to an order in a memory designated as a resource planner and
• - depending on the orders saved in the resource planner Control of the switching operations provided in the corresponding nodes.

After receiving an order, the control circuit checks whether a path with the corresponding quality of service through the network, for example between two nodes or between two subscriber lines connected to different nodes are connected, are possible. The respective nodes or subscriber connections are called endpoints. Such a route or path search is done with the The term "routes" denotes. It is first checked whether those in the path corresponding nodes lying resources, d. H. the necessary transmission and Can provide processing capacity at the required time. A Routed job is stored in a memory that is used as a resource planner referred to as. The available resources depend on the time routed orders. At predetermined times, the in Resource planners allocate resources from the corresponding nodes switched.

The resource planner has entries about the at predetermined times Use of resources. The control circuit stores at least one Order with the resources to be used at the start and end time in this resource planner.  

So that the control circuit learns which resource at a certain point in time the entire order must be placed before the point of interest time period of a resource can be read in the resource planner. Around To avoid this, the control circuit continues to save the orders with the resources to be used between the start and end time Times in the resource planner. This allows the control circuit to determine which resource is to be used at any time.

To simplify the administration or accounting of the resource planner, the control circuit stores a summary of various resources Cluster as order and if a cluster is used, this will be together with the order is saved in the resource planner. It is also possible to form parent clusters or "super clusters". Here, the Control circuit either different clusters or different clusters and Resources are saved as aggregate parent clusters and as one Order handled.

The resource planner can also identify one or more Resources that are down or serviced are being used. Here the control circuit stores failure or maintenance of a resource or of a cluster like an order. So at times they seem to occupy one Resource. The duration of this virtual order is determined by the repair or Maintenance time determined. In the event of a conflict, these virtual orders are even depriving already accepted orders of the resources afterwards, whereby a new routes must be carried out. In the event of failures, however, only then rerouted if the defective resource cannot be repaired in time.

The control circuit with an input module is used to enter orders coupled. Orders entered are then in the control circuit on their  Feasibility checked. An accepted or feasible order will be in cached in a queue.

The control circuit must have a specific strategy when there are several orders apply to route the orders. A first strategy is the path and the nodes with the resources required for a job in turn follow the arrival of the orders to process. Another strategy worked on the orders according to the order of the given start times of the respective Orders or in the order in which the orders arrive. Orders can by the control circuit also according to a predetermined processing priority the profit or according to the order of complexity of the orders will.

The network according to the invention is explained in more detail with reference to the drawings. It demonstrate:

Fig. 1 shows a part of a network having multiple nodes and multiple nodes connected lines,

Figs. 2 and 3 show two different block diagrams, the processes for the treatment of orders over a connection concern,

Fig. 4, 7 and 9 resource directions in matrix form for displaying routed orders with certain resources at certain times,

Fig. 5 shows an event band for displaying routed jobs and

Fig. 6, 8 and 10 extended event-bands to represent routed orders.

In Fig. 1, a part is illustrated of a network having a plurality of nodes 1 to 9. Such a network can be an SDH network (SDH = Synchronous Digital Hierarchy), an ATM network (ATM = Asynchronous Transfer Mode), a PDH network (PDH = Plesiochronous Digital Hierarchy) etc. The nodes could be cross-connects, add-drop multiplexers, multiplexers, etc.

The network contains nodes 1 to 9 , which are connected to each other via different physical lines. The lines can each have different transmission capacities (e.g. 2 Mbit / s or 155 Mbit / s) and can carry one or more signals. For example, a multiplex signal can be transmitted according to the synchronous digital (SDH) or the plesiochronous digital hierarchy (PDH). A connection between two nodes or endpoints with a certain capacity and quality of service is characterized as a logical path. A path can run across several intermediate nodes. Likewise, several logical paths can run between two nodes over a physical line.

To determine the nodes for a path, a route search is carried out, which also includes routes is called, carried out by means of a computer and appropriate software. It must be checked whether the nodes of a path have resources in the form of Can provide transmission and processing capacity. So it must be z. B. it is determined whether a customer receives a Connection between two endpoints can be made possible.

Different connections exist between nodes 1 to 9 . One or more resources can be available via a connection between two nodes. For the sake of simplicity, there is a resource below for each connection between two nodes. For example, the connection between nodes 1 and 2 has a resource R1, the connection between nodes 3 and 4 has a resource R2, the connection between nodes 2 and 4 has a resource R3, the connection between nodes 2 and 5 has a resource R4, the connection between nodes 4 and 6 is a resource R5, the connection between nodes 6 and 9 is a resource R6, the connection between nodes 3 and 7 is a resource R7 and the connection between nodes 7 and 9 is a resource R8.

The routing and control of nodes 1 to 9 is carried out by a control circuit 10 containing network management software, which is indicated by the dashed lines between control circuit 10 and nodes 1 to 9 . Inputs (e.g. control commands) can be made by an operator via an input module 11 which is coupled to the control circuit 10 .

Orders from a customer are entered via the input module 11 , to which the control circuit 10 carries out corresponding control processes. In addition to the customer name, such an order describes the following parameters:

• - connection endpoints
• - Start time and duration of the connection
• - transmission capacity
• - Quality of service.

The connection endpoints indicate, for example, that a connection should be switched between two subscribers A and B or between two nodes. So z. B. the network management software contained in the control circuit 10 (hereinafter referred to as network management) for a desired connection between the nodes 3 and 6 ( FIG. 1), the resources R2 and R5 are available to the customer. In addition to the start time and duration of the connection, the periodicity of a connection can also be specified. The required transmission capacity (e.g. 2 Mbit / s) and the quality of service are also specified.

The procedure for processing customer orders for network management can be explained using the block diagram shown in FIG. 2. A block in Fig. 2 indicates a function or a process of network management. Incoming orders A are first subjected to an acceptance test (block 12 ). It is checked in block 12 whether an order A entered via the input module 11 can be carried out at all. For example, general criteria can be used to test whether the order requested by the customer can be accepted. Such a criterion results, for example, from a utilization statistic or another criterion uses an estimated or approximate utilization for the requested period. The acceptance test does not decide which path is made available to the customer.

After the acceptance test has taken place, either a rejection Z or a confirmation B of the order is issued by block 12 . A confirmation B means that the order can be fulfilled with a high degree of probability. Confirmed orders are placed in a queue (block 13 ). In block 14 , the jobs placed in the queue are routed, ie a path is searched between the two end points and, if the path is found, the corresponding resources are reserved for the requested period. The block 14 is supplied with the resources available for routing from a block 15 containing a memory, which is also referred to as resource planner 15 below. If a path has been found for a job at the desired time, this routed job is stored in the resource planner 15 . Orders that cannot be routed are communicated to block 12 , which then issues a rejection. The resource planner delivers the required switching orders to block 16 at the predetermined times, for example when an order starts or ends, which then executes these in the network according to FIG. 1 (switching). Furthermore, the resource planner 15 can also read the block 12 data for an acceptance test in order to find an approximate overview of the utilization.

FIG. 3 shows an alternative block diagram to that of FIG. 2, which specifies another procedure for processing customer orders from network management. In this procedure, the premature acceptance test provided in FIG. 2 is not carried out. In the block diagram of FIG. 3, incoming orders are first placed in a queue (block 17 ) and then routed (block 18 ). Block 18 also issues a confirmation B or rejection Z of the order. A routed job is stored in a memory contained in block 19 , which is also referred to below as resource planner 19 . The feedback loop in Fig. 3 between blocks 18 and 19 indicates that even routed jobs must be rerouted under certain conditions to find a better global solution. Otherwise, the function of block 19 in FIG. 3 is identical to the function of block 15 in FIG. 2. Block 20 in FIG. 3 has the same function (carry out switching orders) as block 16 in FIG. 2.

After routing, an order is stored in the form of a two-dimensional dynamic matrix in the memory contained in block 15 ( FIG. 2) or in the memory contained in block 19 ( FIG. 3). As already mentioned above, this memory is called the resource planner and specifies for which point in time which resource is reserved for which orders. Such an example resource planner is shown in FIG. 4. The continuous time is quantized and a matrix column corresponds to a unit of time. A time unit can be 30 minutes, for example. The lines correspond to occupied resources. Resources that do not appear in the planner of FIG. 4 are free. As a result, the planner specifies the overall network state for each point in time as required by blocks 16 ( FIG. 2) and 20 ( FIG. 3) for switching.

The planner in Fig. 4 shows six resources R1 to R6. The resource R1 is between the times t2 and t5 with the order E, the resource R2 between the times t1 and t3 with the order D, between the times t4 and t6 with the order A and between the times t7 and t8 with the order C. , the resource R3 between the times t7 and t8 with the job C, the resource R4 between the times t2 and t5 with the job E, the resource R5 between the times t1 and t3 with the job D and between the times t4 and t6 with order A and resource R6 between times t1 and t3 with order D and between times t4 and t6 with order B. The job A thus uses the resources R2 and R5 ( FIG. 1) in order to establish a connection between the nodes 3 and 6 . The order B uses the resource R6 between the nodes 6 and 9 for an assignment, the order C the resource R2 and R3 between the nodes 2 and 3 , the order D the resource R2, R5 and R6 between the nodes 3 and 9 and the Order E the resource R1 and R4 between nodes 1 and 5 .

The planner can be stored in the form of a matrix according to FIG. 4 in the memory represented by block 15 ( FIG. 2) or block 19 ( FIG. 3). A space-saving method for storing the planner is explained below with the aid of FIG. 5. The modularized planner according to FIG. 4 must be transformed into an event tape, which is shown in FIG. 5. At a certain point in time, the event band indicates the start of an order (e.g. start of order A: start A) with the resources used (e.g. for order A (start A): R2, R5) and the end an order (e.g. for the end of order A: end A) with the resources used (R2, R5). At times t1 to t8 there are the following entries in the event band:

t1: start D (R2, R5, R6)
t2: start E (R1, R4)
t3: end D (R2, R5, R6)
t4: start A (R2, R5), start B (R6)
t5: end E (R1, R4)
t6: end A (R2, R5), end B (R6)
t7: start C (R2, R3)
t8: end C (R2, R3).

For example, at time t3, the order D (end D) that contains the resources ends Has used R2, R5 and R6, and the order A with the resources R2 and R5 and order B with resource R6 start at time t4 (start A and start B).

The planner in the form of the event band according to FIG. 5 allows all occupied resources to be determined at any time. However, a simple read access to the corresponding matrix column is not sufficient for this. Rather, the planner must be read from the first time entry until the desired time. It is determined whether a resource is bound or released at the time read. In this way, exactly the resources used result at all times.

In order to avoid reading the event band from the beginning of the first time entry and to make read access in the corresponding matrix column only necessary, the event band according to FIG. 5 is converted into an extended event band according to FIG. 6. This extended event band has redundancy. In addition to the start (start) and end (end) of an order, it also indicates at any time whether an order still exists (e.g. continuation of the order D: cont D). The resources R2, R5 and R6 are also specified here. At times t1 to t8 there are the following entries in the extended event band:

t1: start D (R2, R5, R6)
t2: start E (R1, R4), cont D (R2, R5, R6)
t3: end D (R2, R5, R6), cont E (R1, R4)
t4: start A (R2, R5), start B (R6), cont E (R1, R4)
t5: end E (R1, R4), cont A (R2, R5), cont B (R6)
t6: end A (R2, R5), end B (R6)
t7: start C (R2, R3)
t8: end C (R2, R3).

For example, at time t5 there is no indication of the end of the opening Letters E (end E) with resources R1 and R4 also specify the next exist the order A and B (cont A, cont B) with the resources R2 and R5 or R6 included. So it can be done without reading a number from the beginning find that, for example, at time t2 the resources R1, R4, R2, R5 and R6 or at time t5 the resources R1, R4, R2, R5 and R6 are bound or needed.

To reduce the number of resources to be switched, i. H. to the Simplifying administration or accounting of the resource planner are several Resources that are assumed by the network operator to be mostly used together, united as a cluster. For example different resources of a part of the network to form a cluster to get nabbed. An order that has a path through this part of the network is then preferably assigned to the cluster. There is also further the possibility of individual clusters or clusters and resources at one summarize parent clusters, so that a "super cluster" results.

In Fig. 7, another exemplary section of the resource scheduler is shown. This has two resources R7 and R8, which are permanently occupied by the cluster C1. Cluster C1 is also listed as a resource, as are resources R7 and R8, as can be seen from FIG. 7. Between times t9 and t10, cluster C1 is assigned job F and between times t11 and t12 job G.

The directions of FIG. 7 can also be in an extended event-band to be transformed. This event band, which is shown in FIG. 8, has the following entries:

t9: cont C1 (R7, R8), start F (C1)
t10: cont C1 (R7, R8), end F (C1)
t11: cont C1 (R7, R8), start G (C1)
t12: cont C1 (R7, R8), end G (C1).

At times t9 to t12, resources R7 and R8 are with the cluster C1 occupied (Cont C1). Orders that use this cluster C1 exist between the times t9 and t10 (start F, end F) and between the times t11 and t12 (start G, end G).

The cluster C1 described by means of FIGS. 7 and 8 is permanent. Temporary clusters can also exist. An example of this is shown in FIG. 9, which shows a further section from the resource planner. The temporary cluster S1 occupies the resources R9 and R0 between the times t13 and t14 and between the times t15 and t18. In FIG. 9, the temporary cluster S1 is also recorded as a resource. This has the states "occupied", "free" or "does not exist". Before time t13, between times t14 and t15 and after time t18, cluster S1 does not exist, between times t13 and t14 and t16 and t17, cluster S1 is free and between times t15 and t16, cluster S1 is with the Job H and job I between times t17 and t18. For this purpose, the corresponding extended event band is shown in FIG. 10, which has the following entries:

t13: start S1 (R9, R0)
t14: end S1 (R9, R0)
t15: start S1 (R9, R0), start H (S1)
t16: cont S1 (R9, R0), end H (S1)
t17: cont S1 (R9, R0), start I (S1)
t18: end S1 (R9, R0), end I (S1).

By forming clusters, the administration or accounting of the Planners and also the temporal alignment problem reduced. Under the alignment Problem is the waste of resource capacity to understand when the Total amount of resources are not available at the same time. This Problem can occur if the resources differ on time limited orders are distributed. Periodic is therefore recommended Serve standing orders with clusters.

To create the planner according to FIG. 4, 7 or 9, the event band according to FIG. 5 or the extended event band according to FIG. 6, 8 or 10, the orders must be routed, which is indicated by block 14 ( FIG. 2 ) or block 18 ( Fig. 3) is specified. In order processing, an optimal path is sought, ie all required resources are determined and reserved for the desired period.

If individual paths are searched sequentially, this is called local optimization designated. When determining several overlapping in time Connections, a global optimum is sought, which depends on the order of the Routes (hereinafter referred to as routing order) depends. That can be seen from of the following example.

For this purpose, it is assumed that the orders A, B and D overlap in time - in contrast to that shown in FIG. 4. Orders A and B should start at time t2. If the order D were first routed by the network management, the resources R2, R5 and R6 would be allocated to it. The much more profitable orders A and B would be rejected. If the routing order was reversed, only one, namely the less tolerable order D would be rejected. However, other resources could be made available to order D. Here the resources R7 and R8 (see FIG. 1) would come into question.

Finding a global optimum is very computationally intensive, because all possible Combinations of orders must be calculated. With n Orders n! Options.

Because finding a global optimum is very computationally intensive, another becomes Possibility applied to find a quasi-optimum. Instead of The routing order is systematically determined according to a global optimum heuristic criteria set.

Such a criterion can be, for example, the strategy, the routing order to be determined after the orders have arrived (first come, first serve). Hereby The customer's orders are to be achieved without any preference to edit. Another criterion can be the strategy, the routing order according to the urgency according to the connection start times (most urgent one first). It can happen that already routed orders are new must be routed again. Once an order has been confirmed, it should not be rejected. If no path or Resource is also available after a redirection, the newly entered must Order will be rejected.

When entering customer orders into the input module 11 ( FIG. 1), a processing priority can also be entered. A strategy for determining the routing order can then also be to determine this according to the processing priority (highest priority first). It can also happen that an already routed job is rerouted again. Further criteria for determining a routing order can be that it is determined according to how profitable a connection is for the network operator (most profitable first), or that it is determined according to the complexity of a connection (most complex one first).

If a required resource fails during an existing connection, there is the possibility, for example, that the network management switches over to a specially reserved resource. If, on the other hand, a resource for a confirmed order fails before the connection is established, network management starts a new route. Block 14 ( FIG. 2) or block 18 ( FIG. 3) are notified of resource failures, so that the process represented by block 14 or block 18 has to reroute the existing orders.

The resource planner can also be used to indicate a failure of one or more resources. A failed resource or a cluster is treated in the resource planner like an incoming order (virtual order). The duration of this virtual order is determined by the repair or maintenance time. If, for example, resource R5 in FIG. 1 has failed, this is identified accordingly in the resource planner. Until the failure is remedied, this resource cannot be taken into account when routing from block 14 ( FIG. 2) or block 18 ( FIG. 3). Maintenance of resources or clusters in the resource planner can also be treated like a temporary job.

When optimizing the routing order, a new failure can also occur be treated like an order with the highest priority, for example already routed jobs that use the failed resource are immediately new be routed. In the event of a conflict, the virtual orders can even withdraw the resources that have already been accepted and thus enter them require new routes. In the event of failures, however, it is only rerouted if the defective resource cannot be repaired in time.

Claims (13)

1. Network with a plurality of nodes ( 1 to 9 ), with a plurality of lines connecting the nodes ( 1 to 9 ) and with a control circuit ( 10 ) which

• provision is made for determining a path with the resources contained in nodes ( 1 to 9 ) for an order which is characterized at least by the end points, the start time, the duration and the transmission capacity of a connection,
• - For the time-dependent storage of the resources assigned to an order in a memory designated as resource planner ( 15 , 19 ) and
• - Depending on the orders stored in the resource planner ( 15 , 19 ) for controlling the switching operations in the corresponding nodes ( 1 to 8 ) is provided.

2. Network according to claim 1, characterized in
that the resource planner ( 15 , 19 ) has entries on the use of resources at predetermined times and
that the control circuit ( 10 ) is provided at least for storing the order with the resources to be used at the start and end time in the resource planner ( 15 , 19 ). 3. Network according to claim 2, characterized in that the control circuit ( 10 ) is further provided for storing the order with the resources to be used at times between the start and end time in the resource planner ( 15 , 19 ). 4. Network according to claim 3, characterized in that the control circuit ( 10 ) is provided for storing a different resource to be clustered as a job and for storing a job with the cluster to be used. 5. Network according to claim 4, characterized in that the control circuit ( 10 ) for storing a different cluster or different clusters or resources summarizing parent cluster is provided as an order. 6. Network according to claim 4, characterized in that the control circuit ( 10 ) is provided for storing a failure or maintenance of a resource or a cluster as an order. 7. Network according to one of the preceding claims, characterized in that the control circuit ( 10 )

• - Coupled with an input module ( 11 ) for entering orders and
• - To check the feasibility of an order and to store the accepted orders in a queue ( 13 , 17 ) is provided.

8. Network according to one of the preceding claims, characterized in that the control circuit ( 10 ) is provided in the presence of several orders for determining the path and the node with the necessary resources for an order according to the order of arrival of the orders. 9. Network according to one of the preceding claims, characterized in that the control circuit ( 10 ) is provided in the presence of several orders for determining the path and the node with the required resources for an order in the order of the given start times of the orders. 10. Network according to one of the preceding claims, characterized in that the control circuit ( 10 ) is provided in the presence of several orders for determining the path and the node with the necessary resources for an order according to a predetermined processing priority of the orders. 11. Network according to one of the preceding claims, characterized in that the control circuit ( 10 ) is provided in the presence of several orders for determining the path and the node with the necessary resources for an order after the profit of the orders. 12. Network according to one of the preceding claims, characterized in that the control circuit ( 10 ) is provided in the presence of several orders for determining the path and the node with the necessary resources for an order according to the order of complexity of the orders. 13. Network according to one of the preceding claims, characterized in that the control circuit ( 10 ) in the event of a resource failure for storing the resources associated with the job in question in the resource planner ( 15 , 19 ) and only if the resource is not repaired in good time Determining a new path is at least intended for the job affected by the failure.