WO2009074179A1

WO2009074179A1 - Capacity planning method

Info

Publication number: WO2009074179A1
Application number: PCT/EP2007/063868
Authority: WO
Inventors: Riccardo Martinotti; Diego Caviglia; Giulio Bottari
Original assignee: Telefonaktiebolaget Lm Ericsson (Publ)
Priority date: 2007-12-13
Filing date: 2007-12-13
Publication date: 2009-06-18

Abstract

A capacity planning method for determining resources required in a communication network is provided. The method comprises carrying out network simulation trials to determine network performance in the event of failure of a network resource or resources, each trial relating to a different failure of one or more resources. The method comprises prioritising the order in which the trials are carried out and determining the resources required from the results of the simulation trials. A network building method, a network, a method of sending a signal through a network, a computer program product and a computer based on this invention are also provided.

Description

CAPACITY PLANNING METHOD

TECHNICAL FIELD

This invention relates to a capacity planning method for determining resources required in a communication network.

BACKGROUND

Capacity planning for a network - determining the necessary amount of resources to satisfy a certain traffic matrix given the network topology and the required traffic survivability - is one of the most challenging tasks a network architect is required to perform. Input parameters for a capacity planning method typically include a traffic matrix containing a plurality of entries each describing the number of circuits to be created between a source and a destination or set of destinations. The traffic matrix may also contain any of the following inputs:

• the required protection/restoration scheme, e.g. 1 + 1 protected, Fast ReRoute (FRR), On The Fly, SNCP;

• the required diversity between worker and protection/restoration circuit (link, node or SRLG - shared risk link groups);

• resources (link, node or SRLG) to be included or excluded during the route computation for the circuits; • traffic parameters for the circuit(s) e.g. the bandwidth, CoS (class of service).

The capacity planning method input parameters may also include the physical topology of the network, that is, the connectivity relationship among the network elements. Further input parameters to the capacity planning process may be the required network survivability, that is, the number and type of the failure(s) the network must survive (e.g. double links failure, single SRLG failure, ...) and the order of application of failures (deterministic or stochastic).

Another input parameter for the capacity planning process might be an ordered list of network interfaces to be used during the capacity planning process.

The capacity planning method typically involves simulating the performance of the network given these different input parameters and determining the amount and types of resources required to provide a suitable network. The outcome of the capacity planning process is the number and type of the needed resources. The resulting network can be saved as an inventory report and used to purchase the required material.

In a known capacity planning process, the way the traffic is routed takes into account not only the administrative cost of the links the circuit will traverse but also the actual network state. This is called Traffic Engineering and allows for putting the traffic (routing the traffic to) where the available bandwidth is. Traffic Engineering routing policy takes into account the network state, i.e. the link occupancy (how busy the link is), the type of circuit to be routed (e.g. circuit bandwidth compared to link's bandwidth) and so on. It can be seen that a final network state (port type, link occupancy and circuit path) depends on the circuit routing sequence.

A technical and theoretical problem with the capacity planning process is due to the fact that the final result is dependent on:

• The sequence the traffic matrix is applied, that is, the order the circuits are created in the network • The order the failures are applied in the network

Given the above, the only way to be sure that the capacity planning results are valid for each sequence of both circuit creation and failures is to run the simulation for each possible case that is a NP complex problem. It easy to see that given a traffic matrix with N entries there are N! possible sequences of applying it.

Similarly, given a network with M resources to be failed there are M(M-I), i.e. M² - M, possible combinations.

The traffic matrix problem can be skipped given that what is actually performed when the traffic matrix is applied is Network Engineering (putting the bandwidth where the traffic is) and then there is no need to apply traffic engineering policy during this phase. It can be considered in another development phase. This removes the dependencies of the routing path from the sequence of circuit creation

An already available planning tool (ASTN-Planner) performs a capacity planning method. This tool fails all the resources (link, node or SRLG) one by one, in a "single failure" simulation, or all the possible pairs of resources, in a "double failure" simulation.

There remain scalability problems due to the M² relationship when considering failure of more than one resource. Therefore the existing ASTN-Planner can be slow.

SUMMARY

According to an aspect of the invention, there is provided a capacity planning method for determining resources required in a communication network, the method comprising carrying out network simulation trials to determine network performance in the event of failure of a network resource or resources, each trial relating to a different failure of one or more resources, the method comprising prioritising the order in which the trials are carried out and determining the resources required from the results of the simulation trials.

As a result of prioritising the trials, more desirable trials may be performed before less desirable trials. For example trials that are thought to have more impact on network performance can be carried out before other trials - in this way more relevant trials can be conducted earlier. In some embodiments, less relevant trials may not be carried out at all. For example an ordered list of trials may be produced with most relevant trials higher in the list, and only some (for example a desired number or percentage or those being above (or in other embodiments below) a certain threshold) may be actually carried out. Alternatively these criteria may be used to exclude some trials form the ordered list altogether. Therefore in some embodiments a capacity planning method according to this invention carries out less trials than a prior art method, whilst carrying out the most relevant trials, or the trials thought to be most relevant.

The method may be implemented as a computer program, running on an appropriate computer.

The method may also be implemented as a network building method in which the identified required resources are arranged to be provided for building a network. For example a network may be built or updated on the fly using this invention.

BRIEF DESCRIPTION OF THE DRAWINGS

Figure 1 is a flowchart showing the operation of an embodiment of the invention; Figure 2 is a flowchart showing the operation of an embodiment of the invention;

Figure 3 is a flowchart showing how the prioritisation occurs in one embodiment;

Figure 4 is a schematic representation of prioritised trial lists in an embodiment;

Figure 5 is a flowchart showing operation of the embodiment of Figure 4;

Figure 6 is a schematic representation of a computer program product carrying software according to the invention; and

Figure 7 is a schematic representation of a computer arranged to carry out a method according to the invention.

DETAILED DESCRIPTION

Referring to Figure 1 , in an embodiment of the invention there is provided a capacity planning method 10 for determining resources required in a communication network. At an initial step, the method comprises carrying out 12 in prioritised order network simulation trials to determine network performance in the event of failure of a network resource or resources. Each trial relates to a different failure of one or more resources. A next step comprises determining 14 the resources required from the results of the simulation trials.

As discussed previously, a commonly used algorithm, such as that employed in the software product "ASTN-Planner" is used for existing capacity planning processes in which there is no prioritisation of the order in which trials are carried out - all possible trials are carried out either randomly or in an iterative manner with no concept of prioritisation. The method 10 allows trials to be carried out in a desired order or sequence - this is not the case with existing capacity planning methods. In some embodiments, the order in which trials are carried out is prioritised by taking into account factors identified as having an impact on network performance. In this way, in such embodiments it is possible, if desired, to arrange the order in which trials are carried out so that the trials which have (or are thought to have) most impact upon network performance are carried out before trials which have (or are thought to have) relatively less impact upon network performance.

In some embodiments, for example referring to Figure 2, there is provided a capacity planning method 20 similar to the capacity planning method 10 except that the step of carrying out 22 network simulation trials in prioritised order is only carried out until an abort condition is reached - it is not necessarily carried out for all possible combinations of trials. In some embodiments the abort condition is reached after a predetermined number of trials are carried out. In some embodiments the abort condition is reached after a predetermined number of trials are carried out without resulting in a predetermined number of required resources being identified. The predetermined number of required resources in such embodiments may be 0, 1 , 2, 3... 1000 or any other suitable number for that network and those resources. In some embodiments the predetermined number of trials is calculated as a percentage of the total number of possible trials. In some embodiments the abort condition is reached when a certain proportion (e.g. all or substantially all) of the trials satisfying a particular requirement have been carried out - for example, all of the trials relating to a certain relevant factor - such as an input parameter, e.g. all trials relating to double links failure. Therefore in some embodiments the number of trials carried out is reduced relative to prior art capacity planning methods - since the trials that are carried out are prioritised, the trials that are thought to be most relevant are carried out and those thought to be relatively less relevant are not carried out. Thus the capacity planning of this invention is efficient compared to prior planning processes. In some embodiments a decision can be made based upon the time available for capacity planning and the most efficient trials can be performed within that time constraint.

In some embodiments the network performance relates to characteristics of a circuit or circuits along which the failure occurs and the method comprises prioritising the order by taking into account any one or more of:

^■ bandwidth of circuits affected by the failure

^■ number of circuits affected by the failure ■ distance between failures.

In other embodiments other relevant factors are taken into account.

In another embodiment, referring to Figure 3, a capacity planning method 30 comprises assigning 32 a priority value to each simulation trial. This value may be in the form of a tag such as a binary number tag.

The method 30 further comprises making 34 a trial list of simulation trials - the trial list in this embodiment includes the assigned priority value associated with each trial. The trials may be ordered by their assigned priority value to provide an ordered trial list. In some embodiments the method includes determining if a trial should be excluded from the trials list based upon the factors identified as having an impact on network performance. If it is decided that the trial should be excluded it is not entered on the list. In an alternative embodiment, the trial list may include such trials. In a further embodiment the trial list may include such trials and then the method includes a further step of filtering the list to remove unwanted trials (e.g. remove trials below a certain threshold of relevance - for example, remove the lowest quartile of trials in the list by relevance).

In some embodiments the step of making 34 the list comprises updating an existing trial list.

The method 30 comprises carrying out 36 the trials in order of assigned priority value from the trial list.

In some embodiments the step of making 34 the list is entirely completed before beginning the step of carrying out 36 the trials. In other embodiments, carrying out 36 the trials might begin before making 34 of the list is completed - for example if spare processing capacity is available within the capacity planning system then carrying out the trials might begin before the list is completely made. If most of the list has been made (e.g. more than a certain amount of the list, e.g. more than 50%, then carrying out the trials may begin). The amounts provided are examples only - any other suitable values can be used.

Referring to Figures 4 and 5, in a further embodiment, a single failure ordered list, SFOL, 40 and a double failure ordered list, DFOL 42 are generated in a capacity planning method 50 by assigning 52 priority values to trials relating to single failures and double failures (for resource pairs) respectively.

The SFOL 40 comprises a list of trials relating to failures of single network resources in assigned priority value order (SFl , SF2, SF3, ...).

The DFOL 42 comprises a list of trials relating to failures of pairs of network resources (double failures) in assigned priority value order (DFl, DF2, DF3, ...). The method comprises carrying out 54 trials from the SFOL 40 in order, followed by carrying out 56 trials from the DFOL 42 in order. In other embodiments the trials from the DFOL may be carried out before the SFOL trials. In other embodiments the SFOL trials and the DFOL trials may be carried out simultaneously or substantially simultaneously or in an overlapping manner.

In one particular embodiment, the SFOL list 40 is made by assigning higher priority values to trials having higher bandwidth of circuits affected by the failure, and if the bandwidths are equal, then assigning higher priority values to trials having higher numbers of circuits affected by the failure.

The DFOL list 42 is made by assigning higher priority values to trials having lower distance between the resources of the pair, and if the distances are equal, then assigning higher priority values to trials having higher bandwidth of circuits affected by the failure, and if the bandwidths are equal, then assigning higher priority values to trials having higher numbers of circuits affected by the failure.

In this embodiment, the traffic matrix is applied accordingly with the Network Engineering policy that is traffic is routed taking into account only the administrative costs of the link and not applying any traffic engineering policy.

After each failure trial the network is moved to initial state.

In order to illustrate this embodiment in detail, the following definitions are provided:

• Res is the list of all the M network resources (Node, Link and SRLG) of the network, i.e. Res(i) indicates the i^th resource;

• M is the total number of resources of the network; • N_R: indicates number of circuits that are routed after a failure, i.e. N_R(i) indicates the number of circuits (that are routed after a failure) associated to Res(i);

• B R: indicates bandwidth of circuits that are routed after a failure, i.e. B_R(i) is the indicates the bandwidth of circuits (that are routed after a failure) associated to Res(i);

• N_Hop: indicates the distance between resources - the minimum number of hops (distance) between a couple of resources, i.e. N_Hop(i,j) is indicative of the distance between Res(i) and Res(j); • A V_Hop: indicates the average of the number of hops of circuits in the network;

• SFOL: Single Failure Ordered List is the ordered list of trials 40 (each trial related to a single resource) - as described above, its sequence is followed during the single failure simulation (i.e. SFOL(k) is the k^th failure trial);

• DFOL: Double Failure Ordered List is the ordered list of trials 42 (each trial related to a pair of resources) - as described above, its sequence is followed during the double failure simulation (i.e. OFOL(k) is the k^th failure trial).

In order to formulate the SFOL 40, for each resource Res(i) :

• the bandwidth of circuits that are routed after a failure is counted and put into B_R(i);

• the number of circuits that are routed after a failure is counted and put into N_R(i).

Considering the highest position in the list being the first, the SFOL is built in this way: • the higher B_R(i), the higher the position in the list;

• among resources with the same B R, the higher N_R(i), the higher the position in the list.

The upper criteria can be written as follows: V /,Je [O₅M - I]; IF

{B _R(j) > B _R<J)) OR ((B_R(i) = B_R(J)) AND (N _R(i) > N _ R(J))) THEN

[ SFOL{a) = Res(i); ySFOLφ) = Res(j);

where 0 ≤ a < b < M;

In order to formulate the DFOL 42, for each resource Res(i) :

For each pair of resources Res(i) and Res(j) :

• the minimum number of hop (distance) between Res(i) and Res(j) is counted and put into N_Hop(i,j). For each circuit:

• the average number of hops is counted and put into AV_Hop.

Considering the highest position in the list being the first, the DFOL 42 is built in this way:

• the lower N_Hop(i,j), the higher the position in the list;

• among resources with the same N_Hop(i,j), the higher B_R(i) + B_R(j), the higher the position in the list;

• among resources with the same N_Hop(i,j) and B_R(i)+ B_R(j), the higher N_R(i)+ N_R(j), the higher the position in the list.

The upper criteria can be written as follows: V i, J e [0,M-I] AND V /w,/ze [0,M -1] IF N _ Hopii, j)<N _ Hopim, n) OR

((N _Hop(i,j) = N _Hop{m,n)) AND ^j \(B_R(i) + B_ R(j) >B_ R(m) + B_ R(n))j

OR

^'(N _ Hop(i, J) = N _ Hop(m, n)) AND

(B_R(ϊ) + B_ R(j) = B_ R(m) + B_ R(n)) AND (N_R(i) + N_ R(J) >N_ R(m) + N_ R(n))

THEN

where 0≤a<b<M (A! - 1); In this embodiment, failures that affect resources far from each other more than a threshold percentage of the AV Hop can be considered, with respect the capacity planning, as two single failures. In other words the effects to the traffic are not summed.

Once the DFOL 42 has been filled, the capacity planning engine can scan DFOL 42 and carry out the relevant trials in the network. Due to DFOL filling criteria, the first trials have a higher impact on the network, the following trials have a gradually decreasing impact.

In some embodiments the capacity planning method comprises adding resources in order to ensure the network survivability to the incoming faults. So, with the progress of DFOL faults application, the network is upgraded as needed with new resources. In some embodiments the method comprises ordering, or installing, or both ordering and installing, required resources identified by the capacity planning method. In some embodiments this ordering and/or installing may take place after the capacity planning method is completed or in other embodiments it may take place during the carrying out of trials as and when it is required or desired.

In an embodiment the invention provides a network building method comprising the capacity planning method or methods described herein and further comprising ordering, or installing, or both ordering and installing, required resources identified by the capacity planning method.

As stated above, in this embodiment, the impact of DFOL trials decreases while scanning DFOL itself. As a consequence, also the rate at which new required resources are identified decreases. The number of identified required resources tends to stabilise (i.e. not increase at all or not increase quickly or significantly quickly) allowing the simulation to be aborted before the list is finished. The aborting condition is based on a "sliding window" monitoring of number of added resources.

Even if DFOL is scanned until its end, without meeting the aborting condition, the proposed method provides a computational saving because, in this embodiment, DFOL is a subset of the whole set of possible double faults, for the current type of resource.

Similar considerations are valid for a single failure simulation, considering SFOL instead of DFOL.

Features of any embodiment can be combined with features of any other embodiment to provide a further embodiment of the invention.

In some embodiments the network simulation trials take into account any one or more of a traffic matrix, network topology, and required network survivability.

In some embodiments each different failure comprises a failure of any one or more resources. In other embodiments each different failure comprises failure of the same resources in a different order. In other embodiments each different failure comprises a combination of these.

In some embodiments, the resources comprise nodes, links, shared risk link groups or any combination thereof.

In some embodiments, the network comprises an automatic switched- transport network. This invention also provides a computer program product containing software, which when run on a computer, is arranged to carry out any capacity planning method described herein. Figure 6 shows such a computer program product in the form of a CD 60.

This invention also provides a computer arranged to carry out any capacity planning method described herein. Figure 7 shows such a computer in the form of a processor 70. The processor may be part of a larger computer.

Various modifications may be made to this invention without departing from its scope. Its scope is defined by the claims.

Claims

1. A capacity planning method for determining resources required in a communication network, the method comprising carrying out network simulation trials to determine network performance in the event of failure of a network resource or resources, each trial relating to a different failure of one or more resources, the method comprising prioritising the order in which the trials are carried out and determining the resources required from the results of the simulation trials.

2. The method of claim 1 wherein prioritising the order comprises assigning a priority value to each trial and sequentially carrying out the trials in order of assigned priority value.

3. The method of claim 1 or claim 2 comprising prioritising the order by taking into account factors identified as having an impact on network performance.

4. The method of claim 3 wherein the network performance relates to characteristics of a circuit or circuits along which the failure occurs and the method comprises prioritising the order by taking into account any one or more of:

^■ bandwidth of circuits affected by the failure

^■ number of circuits affected by the failure ■ distance between failures.

5. The method of any preceding claim comprising making or updating a trial list of ordered trials listed in order of assigned priority.

6. The method of claim 5 wherein the trial list is completely made or updated before carrying out the network simulation trials.

7. The method of claim 5 or claim 6 further comprising determining if a trial should be excluded from the ordered trials list based upon the factors identified as having an impact on network performance.

8. The method of claim 3 further comprising excluding trials from the ordered trials list based upon the factors identified as having an impact on network performance.

9. The method of any preceding claim comprising aborting the capacity planning method if an abort condition is reached.

10. The method of claim 9 wherein the abort condition is reached when a predetermined number of trials are carried out without resulting in a predetermined number of required resources being identified.

11. The method of claim 10 wherein the predetermined number of required resources is zero.

12. The method of any preceding claim wherein the network simulation trials take into account any one or more of a traffic matrix, network topology, and required network survivability.

13. The method of any preceding claim wherein each different failure comprises a failure of any one or more resources or failure of the same resources in a different order.

14. The method of any preceding claim comprising carrying out network simulation trials to determine network performance in the event of failure of a single network resources, each trial relating to failure of a different resource, wherein prioritising the order comprises assigning higher priority values to trials having higher bandwidth of circuits affected by the failure, and if the bandwidths are equal, then assigning higher priority values to trials having higher numbers of circuits affected by the failure.

15. The method of any preceding claim comprising carrying out network simulation trials to determine network performance in the event of failure of two network resources, each trial relating to failure of a pair of resources, wherein prioritising the order comprises assigning higher priority values to trials having lower distance between the resources of the pair, and if the distances are equal, then assigning higher priority values to trials having higher bandwidth of circuits affected by the failure, and if the bandwidths are equal, then assigning higher priority values to trials having higher numbers of circuits affected by the failure.

16. The method of any preceding claim wherein the resource comprises a node, a link, a shared risk link group or any combination thereof.

17. The method of any preceding claim wherein the network comprises an automatic switched-transport network.

18. A network building method comprising receiving information relating to required resources determined from the method of any preceding claim and further comprising ordering, or installing, or both ordering and installing, required resources identified by the capacity planning method.

19. The network building method of claim 18 comprising the capacity planning method of any of claims 1 to 17.

20. A network, updated or built using the method of any of claims 1 to 17, or claim 18 or claim 19, or both.

21. A computer program product containing software, which when run on a computer, is arranged to carry out the method of any of claims 1 to 17, or claim 18 or claim 19, or both.

22. A computer arranged to carry out the method of any of claims 1 to 17, or claim 18 or claim 19, or both.

23. A method of sending a signal through a network comprising sending a signal through the network of claim 20.

24. A capacity planning method, network building method, network, computer program product or computer substantially as herein described with reference to any one or more of the accompanying drawings.

To be accompanied, when published, by Figure 1 of the accompanying drawings.