WO2013000711A1

WO2013000711A1 - Methods and apparatuses for creating addresses for subscribers in a network

Info

Publication number: WO2013000711A1
Application number: PCT/EP2012/061215
Authority: WO
Inventors: Hermann Angst; Franz-Josef GÖTZ; Joachim Lohmeyer; Matthias Scheffel
Original assignee: Siemens Aktiengesellschaft
Priority date: 2011-06-30
Filing date: 2012-06-13
Publication date: 2013-01-03
Also published as: DE102011078424A1

Abstract

The invention relates to methods and apparatuses for creating addresses for subscribers in a network such that hash collisions are avoided on account of the allocation of the addresses.

Description

description

Methods and apparatus for creating addresses for subscribers in a network

The invention relates to methods and apparatus for

Creating addresses for participants in a network.

In addition to traditional communication networks, more and more autonomous subscribers, such as network nodes or terminals, exchange data with one another. Thus, in addition to large networks with several hundred participants, a variety of small networks with, for example, up to 100 participants. Addressing of these subscribers takes place, for example, by means of MAC addresses (MAC - Media Access Control) with a 2 ^× 48 address space ( ^Λ means to power, ie 2 ^× 48 is equivalent to 2 × 48), IPv4 addresses having a 2 ^× 32 Address space and IPv6 addresses with a 2 ^Λ 128 address space (IP - Internet Protocol, v4 / v6 - version 4/6).

The respective address of the subscribers is stored at certain points in the network, such as in a switch that controls a distribution of data packets for subpaths according to OSI layer 2 (OSI Open Systems Interconnection).

It is disadvantageous that, especially in large networks with a large number of subscribers, storing the addresses means a considerable amount of memory. In addition, a search for an address with many digits in a forwarding table is expensive.

Therefore, instead of "real" addresses, shortened addresses, which are referred to below as address values, can be used in a forwarding table, for example by means of a hash function [1] to convert the address into the address value, for example, from the address with 48 bits in an address value with 12 bit a reduction in the address length achieves the memory requirement, as well as an acceleration in the search for an address value in the routing table. However, the shortening of the address also requires that different addresses be encoded by the hash algorithm to an identical address value. This is called a hash collision. An approach to this is, when occurring

Hash collisions provide a memory area in the form of a list in which addresses are entered with the same hash value. This does not solve the problem satisfactorily because a number of list entries and thus a permissible number of addresses with the same hash value are still limited. When searching for an address by means of the address value, this results in a considerable amount of time. Another approach is described in [2]. An implementation of a hash function optimized for a certain address distribution in LA switches (LAN Local Area Network) is presented. However, this approach is associated with a considerable implementation effort and is therefore usually excluded for commercial applications.

Therefore, the object is to provide methods and devices which avoid a hash collision and allow a simple realization.

This object is solved by the independent claims. Further developments of the invention can be found in the dependent claims.

The invention relates to a method for creating a first address of a first subscriber in a network with the first subscriber and at least one second subscriber, wherein the second subscriber has a second address, and a length of the addresses is greater than a length of address values associated with the addresses, comprising the steps of: i) providing the second address value of the second subscriber by applying a hash function based on second

Address;

ii) creating the first address from a set of possible addresses such that both the first address is different than the second address and the first address value of the first user generated by applying the hash function based on the first address to the second address value.

This method is advantageous because a complex error treatment in the case of hash collisions can be avoided. In addition, a memory space for storing e.g. Redirection information for address values in routers or switches is reduced because no exceptions for hash collisions need to be stored. In addition, a time to process lists based on address values is reduced because no hash collision routines need to be edited.

The address means an address of a subscriber in a network, e.g. Sender and / or recipient address. A length of the address, e.g. 48bit, is longer than a length of the address value associated with the address, e.g. 12bit. Thus, the address generally has more bits than the address value that can be assigned by means of the hash function. Under Address an IP address according to IPv4 or IPv6 (IP-

Internet Protocol, v4 / v6 version 4/6) or a MAC address (MAC Media Access Control). Further details can be found in the introduction. The hash function is any algorithm that transforms the address into an address value that is shorter in length than the address. A hash function can be a classic hash algorithm, such as known from a document [1].

In addition, the invention can be implemented by a method for creating a first address of a first subscriber for a network, which has at least two further subscribers with associated further addresses, and a length of the addresses is longer than a length of address values belonging to the addresses following the steps below:

Determining two further address values using a hash function based on the two further addresses;

Create, if the two further address values are identical, the first address for the first further subscriber as the first subscriber, taking into account a second address, which is the further address of the second further subscriber, from a set of possible addresses such that both the first address different from the second address, as well as the first address value of the first subscriber, which is generated by applying the hash function based on the first address, is different from a second address value, which is the further address value of the second further subscriber;

Assigning the first address to the first further subscriber if the two further address values are identical.

This variant of the method makes it possible in an existing system to change addresses which cause a hash collision in pairs in such a way that the hash collision can be avoided. The resulting advantages are analogous to the previously described method.

In a first extension of the procedure, the creation of the first address is performed as follows:

a) selecting one of the possible addresses as the first address; b) generating the first address value;

c) comparing the first address value with the second address value; d) repeating steps a) to d) if the first address value is identical to the second address value.

This allows a simple and reliable implementation of the method step "creating the first address".

As an alternative to the first extension, creating the first address in a second extension can be done as follows:

Creating the set of possible addresses such that all address values formed by applying the hash function based on the possible addresses are each pairwise different, and the second address is not part of the set; Selection of one of the possible addresses of the set as the first address.

This allows a simple and reliable realization of the method step "creation of the first address." In addition, this second extension generally achieves a quicker selection than the first extension, since a "trial and error" of candidates for the first address, as in the case of first extension, not applicable. The second extension can be supplemented in an advantageous manner by deleting the first address from the set after selection of one of the possible addresses, whereby the selection is further accelerated, since only unallocated addresses are taken into account in the selection.

The invention further relates to an apparatus for establishing a first address of a first subscriber in a network with the first subscriber and at least one second subscriber, wherein the second subscriber has a second address, and a length of the addresses is longer than a length of the addresses associated with the addresses, by the following means: First means for providing the second address value of the second party by applying a hash function based on the second address;

Second means for selecting the first address from a set of possible addresses such that both the first address is different from the second address and the first address value of the first user generated by applying the hash function based on the first address; is different from the second address value.

The advantages of the device are analogous to those of the corresponding method steps.

The invention also relates to a device for generating a first address of a first subscriber for a network, which has at least two further subscribers with associated further addresses, and the addresses are longer than address values associated with the addresses, by the following means:

First means for determining the two further address values using a hash function based on the two further addresses;

Second means for creating, if the two further address values are identical, the first address for the first further subscriber as first subscriber, taking into account a second address, which is the further address of the second further subscriber, from a set of possible addresses such, that both the first address is different from the second address and the first address value of the first subscriber, by applying the hash function on

Basis of the first address is generated, different from a second address value, which is the further address value of the second further subscriber;

Third means for assigning the first address to the first further subscriber if the two further address values are identical. The advantages of this extension of the device are analogous to those of the corresponding method steps.

In a first extension of one of the devices, the second means for creating the first address is set up in such a way that the following steps are feasible:

c) comparing the first address value with the second address value;

d) repeating steps a) to d) if the first address value is identical to the second address value.

The advantages of this extension of the device are analogous to those of the corresponding method steps.

In a second extension of one of the devices, the second means for creating the first address is set up in such a way that the following steps are feasible:

The advantages of this extension of the device are analogous to those of the corresponding method steps. Advantageously, the second means is further adapted to delete the first address from the set after selecting one of the further addresses. This further speeds up the selection, as only addresses that have not yet been allocated are taken into account in the selection.

Preferably, the method can be used in a switch of a network, wherein the network is in particular an administered network. As a result, in the admi- network switches (= Locally Administered Network according to IEEE 802) avoid switch hash collisions, which enables a more efficient implementation of the switch.

The invention and its developments are explained in more detail with reference to figures. Show it

FIG. 1 shows a network with a first and second subscriber and a device in the form of an assignment unit according to a first exemplary embodiment

Figure 2 operation of a hash function

FIG. 3 shows a network with a first further and second further subscriber according to a further exemplary embodiment. FIG. 4 shows a flowchart of a method and structure of a method

contraption

Elements with the same function and mode of operation are provided in the figures with the same reference numerals.

FIG. 1 shows a network NET with two subscribers T1, T2 and a device DEV, which is also referred to as an allocation unit. The allocation unit and the subscribers can send messages for exchanging data via paths PI, P2, P3. These paths are realized, for example, by means of copper wire lines, whereby data can be exchanged by means of MAC (MAC-Media Access Control) addressing. The assignment unit DEV has the task of receiving a request message from a subscriber to this configuration data, such as to send its MAC address. Such MAC addresses are referred to as addresses in this example. The network NET is, for example, a network with "locally administered addresses" (= locally administered addresses) according to a Standard IEEE 802 (IEEE Institute of Electrical and Electronics Engineers).

The second subscriber T2 has a second address L2 with 01.44.8A. BC .AA.64 on. In the following explanations, all addresses are displayed in hexadecimal notation. When switching on the first subscriber Tl this sends a request message to the assignment unit DEV. The allocation unit selects from a set ML of the possible addresses a first address Li for the first subscriber. The procedure is as follows:

First step Sl:

A second address value K2 of the second subscriber T2 is formed by applying a hash function HASH on the basis of the second address L2. The hash function, see for example document [1], maps the address with 48 bits into an address value with 8 bits. In this example, the hash function discards the first 40 bits of the address and generates a remainder resulting from dividing the last 8 bits of the address by a value of 7. Thus, the second address value K2 is calculated to be 2.

Second step S2:

The assignment unit takes the first address Li from the

Amount ML. The first address Li is EF .01. B2. F9.79.37. The associated first address value Kl results in 6. Thus, both the first and second address and the first and second address value are different, so that a valid first address was determined.

This first address transmits the assignment unit DEV in the form of a reply message to the first subscriber. Thus, only addresses are available in the network whose address value can be uniquely assigned to an address. This avoids a hash collision. The term hash collision is known from the prior art. In a variant of the above example, the allocation unit DEV initially selects the first address Li equal to AI for the first subscriber. B2. C3. D4. E5.41 off. The associated first address value is 2. Since the first and second address value Kl = K2 is identical, the above selection of the first address is discarded and searched for a first address in the set of possible addresses until a first address is found, the associated first address value is different from the second address value.

Alternatively, the allocation unit may set the amount of possible addresses in advance. In this case, only addresses are added to the set ML which lead to address values which are not generated by addresses already contained in the set ML. For the above hash function, a maximum of 7 different address values 0 to 6 can be generated, so that a maximum of 7 addresses can be included in the set ML, as shown in the following list: Address Address value

A0.7F.88.00.FF.07 0

CC.6A. FF. FF.90.78 1

01.44.8A.BC.AA.64 2

A1.B2.C3.D4.E5.42 3

0:00:10. DD.00. C8 4

00.00.FF.00.66.9F 5

EF.01.B2. F9.79.37 6

To simplify the management of this list, already assigned addresses from the set ML can be deleted. Therefore, after assigning the first and second addresses, the following new list results:

Address address value

A0.7F.88.00.FF.07 0

CC.6A. FF. FF.90.78 1

A1.B2.C3.D4.E5.42 3

00.00.10.DD.00.C8 4 00.00.FF.00.66.9F 5

In addition to the subscribers already explained and the allocation unit, FIG. 1 also shows a switch R at the edge of the network. A switch mediates data packets on OSI Layer 2 (OSI Open Systems Interconnection), e.g. in the so-called MAC layer (= MAC layer). The switch determines on which path or partial path of three possible paths PI, P2, P3 a packet of the first subscriber Tl is forwarded. The forwarding decision made by the switch in FIG. 1 relates to the path from the set of paths. In another example, the forwarding decision may involve only a partial path, from the switch to a next-in-place switch or hop. To perform the forwarding decision, the switch takes the destination address as the address from the received packet and reduces it to its associated address value. Internally, the switch has a forwarding table, in which an assignment between address value and path or subpath is stored, which, for example, looks like this:

Address value path / partial path

2 P3

6 PI

Thus, the switch is able to use the address value to determine the further route or partial path of the packet or message. This embodiment shows the invention by way of less

Attendees. For several hundred users with long addresses, e.g. 128 bit, the invention achieves a significant memory reduction in the above forwarding list, which can be efficiently traversed by the address value due to the address value.

FIG. 3 shows an existing network NET, in which two further subscribers TW1, TW2 with already further assigned subscribers Addresses LWl, LW2 with LWl = A0.7F .88.00. FF .07 and LW2 = AB .4F .4E .00.00. OE present. The determination of the two further address values KW1, KW2 belonging to the two further addresses LW1, LW2 shows that both address values KW1 = KW2 = 0. Thus, the switch R can not differentiate the other participants based on their associated address values. Therefore, the allocation unit assigns a new address to one of the other two subscribers, such as, for example, the first further address LWl = Al. B2. C3. D4. E5.42. Here, either the original additional address can be replaced by the new address or the other subscriber receives the new address as an additional address, which he rewrites on receipt of a packet in the original further address or sending a packet with the original further address in the new address changes. Instead of the further subscriber, another unit (not shown in the figure) can also carry out the conversion of the address from the original address to the new address and vice versa. The invention and its developments can be realized by means of a device DEV for creating a first address Li of a first subscriber T1 in a network NET with the first subscriber T1 and at least one second subscriber T2, the second subscriber T2 having a second address L2, and a length of the addresses Li, L2 is longer than a length of address values Kl, K2 associated with the addresses Li, L2, by the following means M1, M2:

First means M1 for providing the second address value K2 of the second subscriber T2 by applying a hash function HASH on the basis of the second address L2;

Second means M2 for selecting the first address Li from a set ML of possible addresses such that both the first address Li is different from the second address L2 and the first address value Kl of the first subscriber T1 obtained by applying the hash function HASH is generated based on the first address Li, which is different from the second address value K2. Furthermore, the device DEV can be configured to create a first address Li of a first subscriber Tl for a network NET, which has at least two further subscribers TW1, TW2 with associated further addresses LW1, LW2, and a length of the addresses Li, L2, LWl, LW2 is longer than a length of the addresses Li, L2, LWl, LW2 associated address values Kl, K2, KWl, KW2 with the following means Ml, M2, M3:

First means M1 for determining the two further address values KW1, KW2 using a hash function HASH on the basis of the two further addresses LW1, LW2;

Second means M2 for creating, if the two further address values KWl, KW2 are identical, the first address Li for the first further subscriber TWl as the first subscriber Tl taking into account a second address L2, which is the further address LW2 of the second further subscriber TW2, of a set ML of possible addresses such that both the first address Li is different from the second address L2 and the first address value K1 of the first subscriber T1 generated by applying the hash function HASH on the basis of the first address Li is, different from a second address value K2, which is the further address value KW2 of the second further subscriber TW2, is;

Third means M3 for assigning the first address Li to the first further subscriber TWl if the two further address values KW1, KW2 are identical.

FIG. 4 shows the sequence of the method with the steps S1 and S2. Furthermore, FIG. 4 also shows the device DEV. This device comprises a first means M1 for

Performing the first step Sl and a second means M2 for carrying out the second step S2, wherein the second means M2 can also be used to carry out extensions of the method of the invention. In addition, the device may have a third means M3, which is designed to associate the first address. The means M1, M2, M3 can be implemented in hardware, software or in a combination of hardware and software. For example, a process sor the steps S1 to S2 stored as program code in a memory attached to the processor and, if necessary, further steps of extensions of the method. Furthermore, the processor has an input and output interface with which the request message can be read in and the first address can be provided to the first user in the form of a response message. In addition, intermediate results of the individual processing steps, such as, for example, the address values K1, K2 or the quantity ML, can additionally be stored in the memory. Furthermore, the means M1, M2 and M3 can also be realized by a fixed wiring of electronic components with a memory module.

The invention and its developments have been explained with reference to several examples. However, the invention is not limited only to the examples. In addition, the exemplary embodiments can also be combined.

literature

Hash Function, Wikipedia article, found on June 21, 2011, htt: / / en. wikipedia. org / wiki / Hash_function

Huntley, C. et al. "Effect of Hash Collisions on the Performance of LAN Switching Devices and Networks," IEEE Conference on Local Computer Networks, ISBN: 1-4244-0418-5, 14-16 Nov. 2006, pages 280-284

Claims

claims

1. A method for creating a first address of a first subscriber (Tl) in a network (NET) with the first subscriber (Tl) and at least one second subscriber (T2), wherein the second subscriber (T2) a second address (L2 ), and a length of the addresses (Li, L2) is longer than a length of address values (K1, K2) associated with the addresses (Li, L2), comprising the following steps:

i) providing the second address value (K2) of the second subscriber (T2) by applying a hash function (HASH) based on the second address (L2);

ii) creating the first address from a set (ML) of possible addresses in such a way that both the first address is different from the second address (L2) and the first address value (Kl) of the first user (T1) passing through Applying the hash function (HASH) based on the first address generated is different than the second address value (K2).

2. Method for creating a first address of a first subscriber (T1) for a network (NET), which has at least two further subscribers (TW1, TW2) with associated further addresses (FO, LW2), and a length of the addresses (Li, L2, LWl, LW2) is longer than a length from to the addresses

(Li, L2, LWl, LW2) associated address values (Kl, K2, KWl, KW2) in which the following steps are performed:

Determining the two further address values (KW1, KW2) using a hash function (HASH) on the basis of the two further addresses (LW1, LW2);

Create, if the two further address values (KW1, KW2) are identical, the first address for the first further subscriber (TW1) as the first subscriber (T1), taking into account a second address (L2) which contains the further address (LW2) of the second another participant (TW2) is from a

Quantity (ML) of possible addresses such that both the first address is different from the second address (L2) and the first address value (Kl) of the first user (T1) generated by applying the hash function (HASH) based on the first address, different from a second address value (K2) which is the further address value (KW2) of the second further user (TW2);

Assigning the first address (Li) to the first further subscriber (TWL) if the two further address values (KW1, KW2) are identical.

3. The method according to claim 1 or 2,

where the first address is created as follows:

a) selecting one of the possible addresses as the first address; b) generating the first address value (Kl);

c) comparing the first address value (Kl) with the second address value (K2);

d) repeating steps a) to d) if the first address value (Kl) is identical to the second address value (K2).

4. The method according to claim 1 or 2,

where the first address is created as follows:

Creating the set (ML) of possible addresses such that all address values formed by applying the hash function based on the possible addresses are each pairwise different, and the second address (L2) is not part of the set (ML) ;

Selection of one of the possible addresses of the set (ML) as the first address.

5. The method of claim 4, wherein

The following steps are performed after selecting the first address:

Delete the first address from the set (ML).

6. Device (DEV) for creating a first address of a first subscriber (Tl) in a network (NET) with the first subscriber (Tl) and at least one second subscriber (T2), wherein the second subscriber (T2) has a second address (L2), and a length of the addresses (Li, L2) is longer than a length of address values (Kl, K2) associated with the addresses (Li, L2) by the following means (M1, M2):

First means (M1) for providing the second address value (K2) of the second user (T2) by applying a hash function (HASH) based on the second address (L2);

Second means (M2) for selecting the first address from a set (ML) of possible addresses such that both the first address is different from the second address (L2) and the first address value (K1) of the first user (T1), generated by applying the hash function (HASH) based on the first address is different from the second address value (K2).

7. Device (DEV) for creating a first address of a first subscriber (Tl) for a network (NET) having at least two further participants (TWl, TW2) with associated further addresses (LWL, LW2), and a length of the addresses (Li, L2, LWl, LW2) is longer than a length of address values belonging to the addresses (Li, L2, LW1, LW2) (Kl, K2, KW1, KW2) by the following means (M1, M2, M3):

First means (M1) for determining the two further address values (KW1, KW2) using a hash function (HASH) on the basis of the two further addresses (LW1, LW2);

Second means (M2) for creating, if the two further address values (KWl, KW2) are identical, the first address for the first further subscriber (TWl) as the first subscriber (Tl), taking into account a second address (L2), the further Address (LW2) of the second further subscriber (TW2) is, from a set (ML) of possible addresses such that both the first address is different from the second address (L2) and the first address value (Kl) of the first Subscriber (T1) generated by applying the hash function (HASH) based on the first address, different from a second address value (K2) which is the further address value (KW2) of the second further subscriber (TW2); Third means (M3) for assigning the first address (Li) to the first further subscriber (TWl) if the two further address values (KW1, KW2) are identical.

8. Device (DEV) according to claim 6 or 7, wherein

the second means (M2) for establishing the first address is set up in such a way that the following steps are feasible:

c) comparing the first address value (Kl) with the second address value (K2);

9. Device (DEV) according to claim 6 or 7, wherein

Creating the set (ML) of possible addresses such that all address values formed by applying the hash function based on the possible addresses are each pairwise different, and the second address (L2) is not part of the set (ML);

Select one of the possible addresses of the set (ML) first

Address .

A device (DEV) according to claim 9, wherein said second means (M2) is further adapted to erase said first address (Li) from said set (ML) upon selection of one of said possible addresses.

11. Use of the method according to one of claims 1 to 5 in a switch of a network (NET), wherein the network (NET) is in particular an administered network.