WO2003077479A1

WO2003077479A1 - Method for the multi-directional exchange of data sets

Info

Publication number: WO2003077479A1
Application number: PCT/EP2003/002265
Authority: WO
Inventors: Wolfram Kress
Original assignee: Wolfram Kress
Priority date: 2002-03-14
Filing date: 2003-03-06
Publication date: 2003-09-18
Also published as: AU2003215634A1; DE10211097A1; DE10211097B4

Abstract

The invention relates to a method for the multi-directional exchange of data sets between connecting devices (20) connected to each other via a network, comprising an automatic transmitting sequence of switch units (10) provided in the network. Said switch units (10) are respectively connected to one or more connection devices and to each other or via hubs (16) or switches (18).

Description

030484wo / KB / we

Process for the multidirectional exchange of data sets

The invention relates to a method for the multidirectional exchange of data sets between connection devices connected to one another via a network, the network in particular being an Ethernet. Furthermore, the invention relates to a network with which the method according to the invention can be carried out, and to a switching device according to the invention arranged in the network for carrying out the method.

Automation systems based on Ethernet are not hierarchical, so that all participants in an automation system via Ethernet, e.g. on the basis of the TCP / IP protocol, can communicate directly with each other. According to the prior art, management functions, such as Program and parameter sets, diagnostic data and high-bandwidth visualization data are transferred to the automation devices, thereby achieving advantages over the fieldbus systems commonly used in industrial automation.

In industrial automation, the control data must be transferred from the source to its destination within the automation network within short and guaranteed times. The response times required in industrial automation are in the lower millisecond range. Such short reaction times are absolutely necessary, for example, for precise control of processing machines, robots, etc. In particular, emergency signals (emergency stops) must be transmitted within short, clearly defined periods. A probability, even if very high, of timely transmission is not acceptable, especially since injuries to operating personnel can possibly be caused.

Such determinable response times, i.e. guaranteed

Due to its collision procedure (Carrier Sense Multiple Access with Collision Detection = CSMA / CD), the Ethernet cannot guarantee maximum reaction times because the Ethernet exchanges messages only with a calculable, statistical and not strictly determinable probability of success. Due to the collision of data records within the Ethernet, long reaction times may not be avoided. Strict determinability of the data exchange cannot be brought about even by so-called switches provided in the network, which have buffers for storing data records. Here, too, it can occur, albeit with a relatively low probability, that a data record is not transmitted within the maximum response time required for industrial automation. (see e.g. Frank J. Furrer; Ethernet-TCP / IP for industrial automation, Hüthig Verlag, 2000).

Further disadvantages when using switches result from their complex configuration and their high prices, which are due to their complex structure. The matrix of a switch requires high-performance microprocessors. The memory requirement for intermediate storage of an incalculably high flood of messages is very large.

The behavior of known automation devices is characterized by a hierarchical behavior. For example, a controller communicates with the sensors and actuators via a fieldbus. The sensors are, so to speak, "asked" for new measured values and only after a request the "answer" is sent back to the control system via the fieldbus. The control system hereby specifies an ordering system that also solves the problem of determinability. Such systems are also referred to as the master / slave principle, which allows response times in control processes According to the state of the art, the automation device itself is responsible for the determinable data transport in a network.

In known Ethernet networks, network infrastructure devices, e.g. Hubs or switches. Hubs are used to connect consumer devices to the network. Their function is essentially limited to signal processing and network arbitration, which ultimately decides which messages are forwarded and which are rejected or rejected. Since even 100 Mbit Ethernet networks lose efficiency from a network load of around 10%, so-called switches have been developed that regulate data traffic in terms of efficiency in a network. In contrast to hubs, no messages are discarded, but rather temporarily stored and forwarded at a later point in time that cannot be determined strictly in time.

The object of the invention is to provide a method for setting up a communication system, preferably based on Ethernet, for industrial automation, which has a time-determinable communication behavior. The response times should preferably be in the lower millisecond range.

The object is achieved according to the invention by a method for the multidirectional exchange of data records according to claim 1 and by a network for carrying out the method according to claim 29 and a switching device for the network for carrying out the method according to claim 30. The method according to the invention for the multidirectional exchange of data records between connection devices or consumers connected to one another via a network has switching devices according to the invention connected to the connection devices. The

Network connection then takes place by connecting the individual switching devices to one another. This connection can be made, for example, directly from the switching device to the switching device or via hubs. According to the method according to the invention, the time-definable data transport takes place through the switching devices. The switching devices are thus constructed according to the invention in such a way that the data are sent from them in a time-determined manner. According to the invention, the time determinability of the network is therefore not, as in the case of fieldbus systems, in the connection or automation device itself, but with the aid of the switching device. The automation devices can thus send data at any time. The determinability required for automation is then realized by the switching devices connected to the connection or automation devices.

The transmission sequence of the individual switching devices is preferably determined automatically according to the method according to the invention. By setting the transmission order, a collision of data records in the network is avoided. The response time of the network can thus be determined in time. For example, depending on the permitted data record lengths, a maximum period in which a data record of a connection device is sent can be determined. Maximum response times can thus be determined. The provision of switching devices according to the invention, which can send data records in an automatically determined transmission sequence, makes it possible in particular to set up the network on an Ethernet basis. According to the invention, the advantages of Ethernet are thus linked to the requirements for industrial automation. The connection devices or consumers connected to the switching devices can continue to transmit data records to the switching devices at any time. According to the invention, the transmission sequence is carried out exclusively with the aid of the intermediate switching devices.

Priorities are preferably assigned to the data records to be sent. According to the invention, the data records with high priority are preferably sent more frequently than data records with lower priority. It is thus possible, for example, for a switching device to prefer data records with high priority as soon as it is their turn to send a data record in accordance with the transmission sequence. High-priority data records can thus overtake lower-priority data records within a switching device. For example, every switching device always sends its existing data record with the highest priority every second transmission process. The number of priorities used and a suitable set of rules regarding when which priorities are preferred can be optimized in particular depending on the system that is operated with the method according to the invention. The maximum permissible system response time must be taken into account.

For the automatic determination of the transmission sequence, it is preferable to assign current transmission authorization numbers for the individual connections of the switching devices or the data records of the individual connection devices. In particular, each connection of the switching device that is connected or can be connected to a connection device or a consumer is thus assigned a transmission authorization number. According to the invention, data records are then always sent in the specified order.

It is particularly preferred to assign the transmission authorization number on the basis of the Ethernet address of the respective connection Switching device or the Ethernet address of the connection device. Such Ethernet addresses are unique and do not appear twice in any network. A hierarchy, for example according to the ascending Ethernet number, can thus be determined in a simple manner on the basis of a comparison of all Ethernet addresses, for example the connections of the switching devices. For this purpose, each switching device preferably sends the Ethernet address of the individual connections or the Ethernet address of the connection devices to all other switching devices that are located in the network. Every switching device is therefore aware of all Ethernet addresses. Each switching device can then create an ascending list of the Ethernet addresses, for example, so that its own transmission location is defined for each individual connection of the switching device and is known to the corresponding switching device.

The switching device required to carry out the method according to the invention is preferably an electronic logic unit which has at least one memory unit and a connection for an ethernet-based connection device and at least one further connection for connection to the network, i.e. with other switching devices or hubs. Switching devices of this type are referred to below as "Quality of Service Switches" (QoS switch).

The method according to the invention is characterized in particular by the fact that by means of the switching devices or QoS switches, both star-shaped and linear automation topologies can be set up without the use of complex configuration tools (software tools), so that the cabling can be optimally adapted to the decentralized automation structure ,

The method according to the invention also makes it possible with hardware and hardware that are significantly smaller than those of commercially available switches Software effort to build a QoS switch. The QoS switch can be a cost-effective single-chip solution with an integrated or externally connected memory chip without the use of complex

Microprocessor architectures are manufactured.

The QoS switch can be directly and cost-effectively integrated into the respective subscriber, eliminating the Ethernet consumer connection or the Ethernet connection device connection. A MAC (Media Access Control) and the physical can then be omitted.

It is also advantageous to integrate several QoS switches by eliminating the corresponding MACs and physicals in a multi-port QoS switch with a corresponding number of consumer connections.

The method according to the invention is characterized in particular by the following advantages:

1. Quality of services are made possible with the prioritized handling of messages, of which more than two message priorities can be supported.

2. The QoS switches can be operated mixed with hubs in a network without losing the properties of the temporal determinability of the message exchange.

3. In addition to the linear topology, the topological network structure also allows star-shaped or mixed structures.

4. The use of the bandwidth of the Ethernet system is still efficient. Message prioritization is used in the described method for faster handling of messages. 5. Higher priority messages can overtake lower priority messages at any time.

6.QoS switches can be used to create strictly determinable network areas that can be linked to other network areas, e.g. Office domains, can also be coupled via QoS switches.

7. Since all messages are direct, i.e. only delayed sending or forwarding to the receiver due to delay times in the logic are significantly better clock balancing accuracies or clock synchronization times than can be achieved in known Ethernet networks.

The invention is explained in more detail below on the basis of preferred embodiments and examples with reference to the accompanying drawings.

Show it:

1 shows a schematic illustration of an example of a network with switching devices provided according to the invention,

2 is a table showing the transmission rights of several participants,

FIGS. 3-6 a schematic representation of several successive transmission cycles,

Fig. 7 is a diagram of the response time depending on different

Priorities,

8 shows a schematic representation of a message cycle and Fig. 9 is a schematic representation of the switching device.

An active ethernet-linked network structure is set up using quality of services switches (QoS switches) 10. Each QoS switch 10 has at least one consumer or connection device connection 12 and at least one network connection 14. The consumer connection 12 is connected to the consumers or connection devices, and the network connection 14 is either connected directly to a network connection 14 of a further QoS -Switches 10 connected or with commercially available network infrastructure components such as Hubs 16. It is only important in this procedure within the network

To use network infrastructure components that also output all incoming messages to all ports. Switches 18 can also be connected via the consumer connections to connect further networks (for example an office network).

In this method, the consumers or connecting devices 20 may only be connected to the consumer connections 12 of the QoS switches 10.

In the example shown in FIG. 1, twenty connection devices 20 are connected via QoS switches 10 and hubs 16. In Fig. 1, both the star-shaped and line-shaped coupling of consumers can be seen. QoS switches 10 can be connected directly between consumer 20 and hub 16 and also allow the wiring to be looped through to the next QoS switch 10.

Each consumer connection 12 of the QoS switches 10 receives the right to send a message all round. The send authorization is taken from a consecutive number that describes the send position. If a QoS switch 10 has several consumer connections 12, a number is assigned to each consumer connection 12 per QoS switch 10. It is advantageous if the numbers of the consumer connections 12 per QoS Switch 10 can be assigned continuously. The consumer connection 12 with the number "1" first receives the right to send a message or a data record. Since all data records pass through all QoS switches 18 and hubs 16, all other consumer connections 12 can observe the flow of messages and count the number of messages sent The QoS switch 10 then sends its consumer connection 12 has the number “2”, etc. If all QoS switches 10 have sent their messages in accordance with their number of consumer connections 12, there is a transmission pause, which marks the end of a transmission cycle. After the end of the transmission break, the QoS switch 10 whose consumer connection has the number “1” has the next transmission cycle. That is, the pending message of the consumer termination 12 with the serial number “1” is sent.

If a consumer connection 12 has no new messages (from the consumer), it still sends a message (dummy) so that the following consumer connections 12 can determine their correct transmission position. It is advantageous if in this case the shortest possible message is sent and the message is not directed to any recipient in this network. In this way, the network is not burdened with unnecessarily long messages and the consumers 20 can be relieved.

A reception memory is assigned to each consumer connection 12, into which all data records sent by the consumer 20 enter. Each consumer connection 12 per QoS switch 10 is assigned an Ethernet reception address (destination) which is identical to the reception address of the connected consumer 20. This can be taken directly from the protocol traffic with the consumer or with the connection device 20. This ensures that every consumer 20 can exchange 20 messages or data records with every other consumer 20. It is advantageous that only messages from the QoS switch 10 are forwarded to the consumers 20, whose receive address matches that of the Message is identical. The system can be expanded in such a way that multicast and broadcast messages can also be transmitted. In this case, a check for multiple addresses (or address spaces) is carried out in the QoS switch 10 per consumer connection 12 and per message.

In general, it can be assumed that messages that are used for control purposes in automation, have short message lengths and often have to be sent in a time-determinable manner. These messages should then be given high priority.

According to IEEE 802.1p, each message can be assigned a priority, which is defined in the IP protocol and supports the Quality of Services (QoS).

During the arrival of the message, the control logic of the QoS switch 10 checks the priority of the message in the IP range of the telegram and then redefines the transmission sequence of all messages or data records that still remain in the QoS switch 10. So it is possible that later received messages with high priority, messages with low priority can "overtake" in the QoS switch.

The QoS switch 10 can be designed or parameterized so that it supports fewer priorities than specified in IEEE 802.1p (7 classes). In this case, all priorities that are less than or equal to III are treated with priority III.

Optionally, maximum message lengths can be assigned to each priority class I to III. This is expedient in automation, since the message lengths for security-relevant messages are generally shorter than pure control messages, which in turn will be shorter than messages that are used for visualization purposes or general management functions and more and more frequently due to classic Internet services (e.g. http server) to be served. In this case, the message length per priority can be limited. If a message exceeds this length, it will automatically be treated with the next lower priority. This option is advantageous in order to obtain the shortest possible response times for high-priority messages and to be able to calculate them in a time-determinable manner.

To set up simple and inexpensive logistics, it is advantageous to operate a quality of service per transmission cycle, with all messages being sent with the highest priority within a first transmission cycle. This means that all messages with priority I are sent by the consumer connections of the QoS switches in a first transmission cycle.

Lower priority messages are then handled in subsequent message cycles. Since longer messages (TCP / IP, http, FTP ...) also have to be sent in Ethernet and they generally do not place such high demands on real-time capability, these messages should be treated with lower priorities. However, in order to be able to guarantee the highest repetition rate for the messages with priority I, only a part of the messages with priority II is sent in the following transmission cycle. However, so that messages with lower priority can also be sent and received in a time-determinable manner, all messages belonging to a priority are guaranteed to be time-determinable in a fixed number of message cycles.

This is explained in more detail below using an example with three message priorities (QoS):

In this example, the message priorities in an Ethernet-based automation network are divided as follows: Ethernet packet length max.

Security-relevant = priority I; 96 bytes

news

Control messages = priority II; 256 bytes

General messages = priority III; 1526 bytes

In this example, all priority I messages receive a send right in the first send cycle. In the next transmission cycle, the half receive all messages of priority II and in a third transmission cycle ¹ A of all messages of priority III transmission rights, or to be transmitted.

The control of the transmission rights per consumer connection can be found in the table in FIG. 2.

The table shows 12 message cycles, each separated by a pause to synchronize the entire system. After the end of the 12th transmission cycle, transmission cycle 1 is restarted after a pause. Message priorities I to III are handled in three successive message cycles.

Each consumer connection evaluates the two lower bits of its send authorization number. The message cycles to which the treatment of message priorities from I to III corresponds in each case are carried along by the QoS switches 10 via a counter per connection device connection 12. By evaluating the table, it is determined whether in the respective transmission cycle, taking into account one's own number, there is authorization to send the correspondingly prioritized message. Logical 0 means in the table that there is no authorization to send and logical 1 means that a data record is sent. In Figs. 3-6 are shown according to the table (Fig. 2) transmission cycles with 12 connection devices and 3 message priorities (QoS). The messages with priority I, ie the highest priority, are hatched, the messages with priority II are dotted and the messages with priority III are shown in white. For illustration purposes, a 1st, a 2nd and a 3rd transmission cycle is shown as a circle, the first transmission cycle being the transmission cycle 22, the second transmission cycle 24 and the third transmission cycle 26.

In a first round (Fig. 3) in the first transmission cycle 22 of each consumer connection, i.e. the consumer connections with the numbers 1- 12, a message with priority I sent. After a pause 28, data records are sent to every second connection device, i.e. of the connection devices with the numbers 1, 3, 5, 7, 9 and 11. In this case, data records of priority II are sent in the second transmission cycle 24. After a further pause 30, 26 data records of priority III become one quarter in the third transmission cycle the connection devices, or a quarter of the consumer connections, in the illustrated embodiment of the connections with the numbers 4, 8 and 12, sent.

After a further pause 32, the messages of the highest priority are again sent in the first transmission cycle 22 (FIG. 4). Data records from all connecting devices 1-12 are sent again. After the pause 34, 24 data records of half of the devices are sent in the second transmission cycle. These are data records of those devices from which no data records were sent in the first round (FIG. 3), ie devices 2, 4, 6, 8, 10 and 12. After a further pause 36, data records are sent in the third transmission cycle 26, data records of another quarter of the devices, in the illustrated embodiment of devices 1, 5 and 9, being forwarded. In the next rounds (FIGS. 5 and 6), data records with the highest priority are sent in the first transmission cycles 22, data records of all connecting devices being sent. The subsequent transmission cycles 24 correspond to the transmission cycles as shown in FIGS. 3 and 4 are shown. The third transmission cycle 26 contains data records of devices 2, 6, 8 in the cycle shown in FIG. 5 and data records of devices 3, 7, 11 in the next cycle (FIG. 6). Thus, after four cycles (FIGS. 3-6) four records with the highest priority of all connecting devices have been sent. Priority II records have been sent twice in this total circulation. Priority III records were sent once per device after the four rounds (Fig. 3-6). With this method according to the invention, an extremely short response time can be achieved, particularly in the case of data records with a high priority.

In principle, this method can be extended to the treatment of further priorities and there are also other options for load limitation of lower message priorities.

With this method, a determinable transmission time for a data set can be determined with the maximum known data lengths, the baud rate, the interframe gap, the number of participants that have to be rounded up evenly. The transmission time and thus the reaction time is determined by the following formulas:

with: _x = priority ( _x = i to)

R _x = response time in usec (worst case) with priority x

T _x = time of a telegram in usec with 100 Mbit Fast Ethernet with priority x D _x = maximum length of the respective Ethernet protocol in bytes

T | = Time of the interframe gap (approx. 1 usec)

n = number of consumer connections operated rounded up to an even number

P = pause (in usec)

follows: T _x = D _x x 8 / 100,000 + T, (usec)

RI = n x TI + n / 2 x TU + n / 4 x Till + 3 x P RII = 2 x RI RIII = 4 x RI

With the formula given above, in the 3-6 shown example determine the guaranteed response times:

With: P = 10 usec; D _Ϊ = 96 bytes; D _π = 256 bytes; D _m = 1526 bytes = max .; the graph shown in FIG. 7 follows. In the diagram shown, the dashed line shows the maximum response time depending on the number of connection devices for messages of the highest priority, ie priority I. For messages of priority II, the response time is by a solid line and for messages of priority III by a dash-dotted line shown.

Each QoS switch 10 has its own unique Ethernet address per consumer connection 12. These are used for diagnostic and management purposes and are used according to the invention for determining the authorization number. The send authorization number is synonymous with the send position of the consumer connection. The Send authorization number is determined from all Ethernet addresses of the consumer connections involved in the network. So z. For example, the consumer connection with the lowest Ethernet address has the send authorization number 1. The consumer connection with the next higher Ethernet address has the number 2 etc. The control logic of each consumer connection must have knowledge of all Ethernet addresses of the consumer connections in the network. The mutual exchange of all Ethernet addresses of the consumer connections takes place by sending the own Ethernet address to all other QoS switches 10. The pauses after the message cycles are used to exchange these addresses. A typical transmission cycle is shown in FIG. At the end of each transmission cycle there is a pause consisting of P1 and P2.

If no messages are received within the breaks P1 and P2, the consumer connection with the authorization number 1 starts the next transmission cycle. Within a send cycle, only messages from consumer connections that have already been registered are sent.

If a new QoS switch 10 is now added to the network, it observes a transmission cycle and detects its end when no more messages have been received within a pause time Pl. To register, this QoS switch sends the Ethernet address of one of its consumer connections. Since all already registered QoS switches also monitor the pause times, all QoS switches receive this address within the pause time P2. As a result, all other QoS switches now send their Ethernet addresses corresponding to the consumer connections. This communication can take place according to the normal Ethernet rules (CSMA / CD), since no determinable real-time behavior of the network is expected in the case of the network configuration. If all Ethernet addresses have been exchanged once, a new transmission authorization number is determined for each consumer connection. This can e.g. B. after the detected pause time Pl. To When the pause time P2 expires, the consumer connection with the lowest Ethernet address starts the next transmission cycle. If a message is received again before the pause time P2 expires, then all Ethernet addresses must be exchanged again.

8 also shows a registration cycle in which no subscriber has yet registered in the network. This condition can e.g. For example, for the sake of simplicity, it is assumed below that only QoS switches with one consumer connection each register. The first participant in the network will not be able to detect a transmission cycle. After a pause P3 that is greater than P1 + If the participant does not receive any messages within the pause time P3, he sends his address again etc. If a second QoS switch is now connected, this will receive a message from the QoS switch that was switched on first Pl starts the second connected QoS switch with the sending of its Ethernet address, which in turn triggers the procedure described above.

8 also shows how further QoS switches integrate into the network at different switch-on times. With this registration procedure it is possible that network sections can independently reconfigure themselves after one or more participants fail. The sending order may change under certain circumstances. In any case, however, complete message cycles are always built up, which guarantee strict scheduling.

The method can alternatively be changed such that only one Ethernet address is assigned to each QoS switch. In this case, all consumer connections must get by with an Ethernet address. During the registration process, all consumer connections per QoS switch then send the same Ethernet address. In this case, the send position of the consumer connections within a QoS switch can be changed in the QoS switch Software or hardware logic can be specified. All other QoS switches or their consumer connections treat the same received Ethernet addresses as in the procedure described above. The determination of your own send position is not affected as long as each consumer connection sends an Ethernet address and thus registers a send slot.

A sufficient common receive memory area is assigned to all consumer connections or connection device connections 12 of a QoS switch 10. E.g. 8 times 1526 bytes (1526 bytes = max. Ethernet frame) for 8 consecutive messages from the consumer and an additional 1526 bytes for a message from the network for temporary storage, if this cannot be forwarded directly to the consumer. The reason for this can e.g. B. be that the consumer wants to send a message and has occupied the line to the QoS switch in half duplex mode. A memory of 9 x 1526 bytes = 13734 bytes is reserved for each consumer connection. With a commercially available memory module of 64k x 16 bit = 128 Kbyte, a sufficient buffer memory for 8 consumer connections per QoS switch is provided.

The QoS switch shown as an example in FIG. 9 consists of two network connections 40, 42 with which the connection to further QoS switches 10 or commercially available hubs 16 (FIG. 1) can be established and a consumer connection 44 to which a consumer is connected can be. The QoS switch consists of the MACs (Media Access Control) and the physicals for generating the signals, an external memory 46 for intermediate storage of messages, a microcontroller 48 for managing the QoS switch and a control logic 50 which is designed as a single-chip solution.

The following are integrated in the chip: Repeater 52 for forwarding and processing the Ethernet protocols, input and output buffers 54, 56, which serve as send and receive buffers and the logic part 50, which controls the transmission rights and message prioritization in the QoS switch 10. All necessary functions and drivers for connecting an external memory 46 and microprocessor 48 are also accommodated in the chip.

Telegrams arriving from the network are checked for their destination address (or address space) and, if necessary, are forwarded to the consumer directly or, if the line to the consumer is busy, indirectly via intermediate storage. To evaluate the destination address, at least part of the telegram is buffered in the backbone input buffer. In parallel, the incoming telegram is sent via the repeater, which only processes the signal, to the other network connection with only a slight time delay, which is only through. Term of the components is determined, issued again. This is important so that time synchronization can be carried out with high accuracy in the network.

Telegrams arriving from the consumer are examined in the device input buffer for their priority in the IP area of the protocol and stored accordingly in the external memory until they have been output via the network connections with the corresponding next free send slot.

The control logic of the QoS switch can be designed in such a way that, in the event of an impending memory overflow, it rejects lower-priority messages in favor of higher-priority messages.

The microprocessor is used to support the calculation of the transmission authorization numbers and the network diagnosis. It is also advantageous to implement an http server that enables external communication with standard software tools, such as. B. Internet Explorer supported. Depending on the number of consumer connections, it may be advantageous to integrate the memory into the chip.

In the event that several consumer connections are to be operated on one QoS switch, only the part of the logic that is necessary for controlling a consumer connection and the corresponding buffers need to be executed multiple times in the chip.

Claims

claims

1. A method for the multidirectional exchange of data records between connection devices (20) connected to one another via a network, in particular based on Ethernet, wherein the time-determinable data transport by switching devices (10) via the connection devices (20) are connected to one another.

2. The method according to claim 1, in which the switching devices (10) are connected to one another without topological restrictions and a transmission sequence of the switching devices (10) is automatically determined.

3. The method according to claim 1 or 2, in which each record to be sent is assigned a priority and records with high priority are sent more often.

4. The method according to claim 3, in which data records with high priority are sent in a first transmission cycle (22) in the specified transmission order and data records with low priority are sent in one or more further transmission cycles (24, 26).

5. The method according to claim 4, in which only data records of the first priority are sent in the first transmission cycle (22).

6. The method according to claim 4 or 5, in which in the first transmission cycle (22) the data records of all connection devices (20) are sent.

7. The method according to any one of claims 1-6, in which data records of part of the connecting devices (20) are sent in the further transmission cycles (24, 26).

8. The method according to any one of claims 1-7, in which only data records of the second priority are sent in a second transmission cycle (24).

9. The method according to any one of claims 1-8, in which in a second transmission cycle (24) data records from about half of the connecting devices (20) are sent.

10. The method according to any one of claims 1-9, in which only data records of the third and / or a higher priority are sent in a third transmission cycle (26).

11. The method according to any one of claims 1-10, in which in a third transmission cycle (26) data records from about a quarter of the connection devices (20) are sent.

12. The method according to any one of claims 1-11, in which the first transmission cycle (22) is carried out after each further transmission cycle (24, 26).

13. The method according to any one of claims 1-11, wherein the transmission cycles (22,24,26) are always performed in order.

14. The method according to any one of claims 8-13, in which in the second transmission cycle (24) data sets are alternately sent from one and the other half of the connection devices (20).

15. The method according to any one of claims 10-14, in which data records of a different quarter of the connecting devices (20) are sent in four successive third transmission cycles (26).

16. The method according to any one of claims 1-15, in which data records of the individual priorities are each limited to a maximum length.

17. The method according to claim 16, wherein when the maximum length is exceeded, a data record is automatically classified in the next lower priority.

18. The method according to any one of claims 1-17, in which the priorities of the data records are described according to IEEE 802.1.

19. The method according to any one of claims 1-18, wherein the determination of the transmission order is carried out by assigning current transmission authorization numbers for the connections (14) of the switching devices (10) or the connection devices (20).

20. The method according to claim 19, in which the assignment of the transmission authorization number of each connection (14) or connection device (20) takes place on the basis of the Ethemet address of the respective connection (14) or connection device (20).

21. The method according to any one of claims 1-19, in which to determine the transmission order of each switching device (10) each has an address, in particular an Ethernet address, of the connections (14) of the switching device (10) or the connection devices connected to it ( 20) is sent to all switching devices (10).

22. The method according to claim 21, wherein each switching device (10) on the basis of the received connection addresses or Connection device addresses and the at least one own connection address or connection device address determine the send position of the switching device (s) connected to it or the connections of the switching devices (10).

23. The method according to claim 19 or 20, in which each switching device (10) counts the number of data records sent by other switching devices (10) in order to determine its own transmission time.

24. The method according to any one of claims 1-22, in which a dummy is sent if there is no corresponding priority record for sending.

25. The method according to any one of claims 21-24, in which the addresses are transmitted in transmission pauses provided between transmission cycles.

26. The method according to any one of claims 21-25, in which the address of a newly connected connection device (20) is transmitted after a first pause time (Pl) and then all addresses for determining the new transmission sequence are transmitted.

27. The method according to any one of claims 1-26, wherein the switching devices (10) forward a data record only to a connection for a connection device (20) that matches the address of the connection device (20).

28. The method according to any one of claims 21-27, in which only switching devices (10) with a connected connecting device (20) send one or more addresses for determining the transmission sequence.

29. Network, in particular based on Ethernet, with a plurality of switching devices (10) connected to one another via network cables, each of which can be connected to at least one connection device (20), the switching devices (10) being constructed such that they are used to carry out the method one of claims 1-28 are suitable.

30. Switching device for networks, in particular based on Ethernet, with

at least one connection for a connection device (20),

at least one network connection (40, 42),

an intermediate buffer for temporarily storing data record to be sent to the at least one connection device (20) or received from the at least one connection device (20)

a logic module which is suitable for communication with logic modules of further switching devices (10) arranged in the network for carrying out the method according to one of claims 1-28.