|
Claims  |
|
|
What is claimed is:
1. A switch for transferring data packets, each data packet having a label
including a first part and a second part, comprising:
pluralities of input ports and output ports, wherein each input port has a
list of internal labels including a first sublist and a second sublist;
and
a switch core for transferring data packets that have arrived at an input
port to a selected output port;
wherein each input port performs, for a data packet that has arrived at the
input port, a first table look-up in the first sublist based on the first
part of the label of the arrived data packet, whereby an internal label is
obtained, to determine whether a second table look-up should be performed;
when the second table look-up should not be performed, the input port
substitutes the internal, label obtained by the first table look-up for
the first part of the label of the arrived data packet; and when the
second table look-up should be performed, the input port performs the
second table look-up in the second sublist based on the second part of the
label of the arrived data packet, whereby an internal label is obtained,
and substitutes the internal label obtained by the second table look-up
for the label of the arrived data packet.
2. The switch of claim 1, wherein, in performing the second table look-up,
the input port uses the first part of the label of the arrived data packet
as an address.
3. The switch of claim 1, wherein the input port performs the first table
look-up by:
forming a combination of the first part of the label of the arrived data
packet and a first base address;
using the combination as an address of an internal label in the first
sublist; and
accessing the internal label corresponding to the address in the first
sublist.
4. The switch of claim 1, wherein the input post performs the second table
look-up by:
determining a pointer from information obtained by the first table look-up
and from the second part of the label of the arrived data packet; and
accessing, in the second sublist, an internal label corresponding to the
pointer.
5. The switch of claim 1, wherein the input port performs the second table
look-up by:
forming a combination of the second part of the label of the arrived data
packet and a second base address obtained from the first table look-up;
using the combination as an address of an internal label in the second
sublist; and
accessing the internal label corresponding to the address in the second
sublist.
6. The switch of claim 1, wherein each output port substitutes a forwarding
label for the label of a data packet received by that output port.
7. The switch of claim 6, wherein each output port has a list of forwarding
labels used for data packets to be forwarded by a respective output port,
the list including a third sublist and a fourth sublist; each output port
performs, when only the first part of the label of a data packet to be
forwarded has been replaced, a third table look-up in the third sublist
based on the first part of the internal label of the data packet to obtain
a forwarding label, and substitutes that forwarding label for the first
part of the label of the data packet; and each output port performs, when
the label of the data packet to be forwarded has been replaced, a fourth
table look-up in the fourth sublist based on the internal label of the
data packet to obtain a forwarding label, and substitutes that forwarding
label for the internal label of the data packet.
8. The switch of claim 7, wherein the output port performs the third table
look-up by:
forming a combination of the first part of the internal label of the data
packet and a third base address;
using the combination as an address for a forwarding label in the third
sublist; and
accessing the forwarding label corresponding to the address in the third
sublist.
9. The switch of claim 7, wherein the output port performs the fourth table
look-up by:
determining a pointer from information obtained by the third table look-up
and from the second part of the internal label of the data packet; and
accessing, in the fourth sublist, a forwarding label corresponding to the
pointer.
10. The switch of claim 7, wherein the output port performs the fourth
table look-up by:
forming a combination of the second part of the internal label of the data
packet to be forwarded and a third base address obtained from the third
table look-up;
using the combination as an address of a forwarding label in the fourth
sublist; and
accessing the forwarding label corresponding to the address in the fourth
sublist.
11. The switch of claim 7, wherein the output port performs the fourth
table look-up by:
forming a combination of the internal label of the data packet to be
forwarded and a fourth base address;
using the combination as an address of a forwarding label in the fourth
sublist; and
accessing the forwarding label corresponding to the address in the fourth
sublist.
12. A network for transferring data packets belonging to logical
connections having either a first type or a second type, each data packet
having a label including a first part and a second part and identifying
the logical connection to which a data packet belongs, comprising:
a plurality of end stations, each end station being either a start point or
an end point for a data packet;
a plurality of switch units, each switch unit having a plurality of input
ports and output ports allows transferring of a, between which ports a
data packet; and
a connection line from each end station to one of an input port and an
output port of a respective switch unit associated with each end station,
and a plurality of connection lines between the input ports and output
ports of the switch units, which are not connected to an end station;
wherein each switch unit establishes, when a data packet arrives from an
end station associated with that switch unit, a logical connection to
another end station, logical connection being established through that
switch unit and in the network in cooperation with other switch units, and
determines an internal label having a first part and a second part for the
logical connection;
each input port has a list of internal labels including a first sublist and
a second sublist and substitutes an internal label for the label of a data
packet arrived at that input port on a connection line which is
established for the logical connection to which the data packet belongs;
when a data packet belongs to the first type of logical connection, the
input port at which the data packet arrives performs a first table look-up
in the first sublist based on the first part of the label of the arrived
data packet to obtain an internal label, and substitutes that internal
label for only the first part of the label of the arrived data packet; and
when a data packet belongs to the second type of logical connection, the
input port at which the data packet arrives performs a second table
look-up in the second sublist based on the second part of the label of the
arrived data packet to obtain an internal label, and substitutes that
internal label for the label of the arrived data packet.
13. The network of claim 12, wherein the first sublist in an input port of
a switch unit comprises information relating to the type of logical
connection and that input port performs, for all data packets arrived at
that input port, the first table look-ups based on the first part of the
labels of the arrived data packets to obtain the types of logical
connection to which the arrived data packets belong.
14. The network of claim 12, wherein an input port of a switch unit inserts
information regarding the type of logical connection in a data packet
arrived at that input port.
15. The network of claim 14, wherein an output port in a switch unit
removes the information regarding the type of logical connection from a
data packet to be forwarded by the switch unit removing the information.
16. The network of claim 12, wherein an input port in a switch unit
performs the first table look-up by:
forming a combination of the first part of the label of an arrived data
packet and a first base address;
using the combination as an address of an internal label in the first
sublist; and
accessing the internal label corresponding to the address in the first
sublist.
17. The network of claim 13, wherein, when a data packet belongs to a
logical connection of the second type, the input port in the switch unit
at which the data packet arrives performs the second table look-up by:
determining a pointer from information obtained by the first table look-up
and from the second part of the label of the arrived data packet; and
accessing, in the second sublist, an internal label corresponding to the
pointer.
18. The network of claim 13, wherein, when a data packet belongs to a
logical connection of the second type, the input port in the switch unit
at which the data packet arrives performs the second table look-up by:
forming a combination of the second part of the label of the arrived data
packet and a second base address obtained from the first table look-up;
using the combination as an address of an internal label in the second
sublist; and
accessing the internal label corresponding to the address in the second
sublist.
19. The network of claim 12, wherein each switch unit, in establishing a
logical connection, determines a forwarding label for the established
connection, and an output port in a switch unit, from which port a data
packet is to be forwarded, substitutes the determined forwarding label for
the logical connection to which the data packet belongs for the internal
label of the data packet.
20. The network of claim 19, wherein each output port has a list of
forwarding labels used for data packets to be forwarded by the respective
output port, the list including a third sublist and a fourth sublist; each
output port performs, when the data packet to be forwarded belongs to the
first type of logical connection, a third table look-up in the third
sublist based on the first part of the internal label of the data packet
to obtain a forwarding label, and substitutes that forwarding label for
the first part of the label of the data packet; and each output port
performs, when the data packet to be forwarded belongs to the second type
of logical connection, a fourth table look-up in the fourth sublist based
on the internal label of the data packet to obtain a forwarding label, and
substitutes that forwarding label for the internal label of the data
packet.
21. The network of claim 20, wherein the output port performs the third
table look-up by:
forming a combination of the first part of the internal label of the data
packet and a third base address;
using the combination as an address for a forwarding label in the third
sublist; and
accessing the forwarding label corresponding to the address in the third
sublist.
22. The network of claim 20, wherein the output port performs the fourth
table look-up by:
determining a pointer from information obtained by the third table look-up
and from the second part of the internal label of the data packet; and
accessing, in the fourth sublist, a forwarding label corresponding to the
pointer.
23. The network of claim 20, wherein the output port performs the fourth
table look-up by:
forming a combination of the second part of the internal label of the data
packet to be forwarded and a third base address obtained from the third
table look-up;
using the combination as an address of a forwarding label in the fourth
sublist; and
accessing the forwarding label corresponding to the address in the fourth
sublist.
24. The network of claim 20, wherein the output port performs the fourth
table look-up by:
forming a combination of the internal label of the data packet to be
forwarded and a fourth base address;
using the combination as an address of a forwarding label in the fourth
sublist; and
accessing the forwarding label corresponding to the address in the fourth
sublist.
25. The network of claim 12, wherein data packets that either arrive
directly from an end station or are forwarded directly to an end station
have a first label format and data packets that are forwarded between
switch units have a second label format;
each input port includes an indicator element for each logical connection
having data packets passing through the respective input port, the
indicator element indicating whether a data packet arriving at the input
port, after being transferred through the respective switch unit, is to be
forwarded from the switch unit directly to an end station or not; and
when a data packet arrives at an input port, that input port decides, based
on the indicator element for the logical connection to which that data
packet belongs, whether that data packet is to be forwarded directly from
the switch unit to an end station;
when the data packet is to be forwarded directly from the switch unit to an
end station, that input port substitutes an internal label having the
first format for the label of the data packet; and
when the data packet is not to be forwarded directly from the switch unit
to an end station, that input port substitutes an internal label having
the second format for the label of the data packet.
26. The network of claim 12, wherein at least one list in a port contains
sequentially all possible contents of label fields which a data packet
received by that port can have, and that port determines either the
internal label or the forwarding label which is to be substituted for the
label of a data packet received by the port from one of the label's
position and sequential number in the list.
27. The network of claim 12, wherein upon an arrival of a data packet at an
input port of a switch unit, that input port adds to the data packet an
extra field, in which an internal address of the data packet is inserted
indicating how the data packet is to be transferred inside the switch
unit, and each output port of the switch unit is adapted to subtract the
extra field from the data packet when the data packet is forwarded from
the switch unit. |
|
|
|
|
Claims  |
|
|
Description  |
|
|
BACKGROUND
The present invention is related to networks and switches for the
transmission or transfer of data packets or data cells and in particular a
processing performed at the incoming and outgoing sides of a switch
involving a translation of the labels, which data packets carry for
identifying that they belong to a particular logical connection
established through the network and the switch.
This application claims priority from Swedish Patent Application No.
9300793-8, filed Mar. 10, 1993, which is incorporated here by reference.
In a data packet network it is required, in the transfer of data packets
from sending units to receiving units through various nodes in the
network, that each data packet can be uniquely assigned a connection,
which has been established earlier in the setup of the logical connection
through the network. It is achieved by providing the data packets with
labels, i.e. a special field comprising identification information. In the
transfer in the network through many nodes, where the data packet is
switched from an incoming connection line to an outgoing connection line,
it may occur that data packets have been provided with the same label what
will cause difficulties or even errors in the transmission.
Data packets belonging to different types of established connections may in
addition have label fields of different formats, e.g. such that data
packets belonging to some connections have a shorter label field while
data packets belonging to other connections also have such a short label
field but also in addition thereto a longer label field. In the second
case the first short label field may then indicate a group of connections,
of which the connection is a part. Such different parts may be present
when for instance some established connections are assigned more permanent
paths in the network than other connections. These different formats
result in further problems in the processing of the labels of the
individual data packets.
The data packets considered here are in the common way transmitted through
the network from one end station to another end station. The end stations
are then provided with a single connection interface, both for sending and
for receiving, coupled through a connection line to an associated
connection node in the network. Special format restrictions may then be
valid for the labels depending if a data packet is transmitted directly
from such an end station or is received by an end station or if a data
packet is in the process of being forwarded between two switching nodes in
the system.
The handling of in particular addresses in data packets is disclosed e.g.
U.S. Pat. Nos. 4,894,822; No. 4,218,756; No. 3,979,733; and No. 4,494,230;
and in European Patent Applications EP-A2 0 282 197 and EP-A2 0 206 403,
and in Soviet Patent Application SU-A 478 445.
In the European patent application EP-A2 0 406 842 a packet switch network
is disclosed comprising label processing at incoming and outgoing links of
a switch. Therein the numbers in data packets identifying the logical
connection--VP or VC number--are substituted by new numbers found in lists
stored in memories at the incoming and outgoing sides of the switch. When
many logical connections are to be simultaneously active, the lists will
be correspondingly long and memory space therefor must be provided in the
construction of the switch.
Label processing is also disclosed in European Patent Application EP-A2 0
481 447, and in U.S. Pat. Nos. 4,947,388; No. 5,099,475; No. 4,995,032;
and No. 5,166,926.
SUMMARY
It is an object of the invention to provide a network and a packet switch
allowing an efficient use of the memory space available for storing label
conversion lists at the incoming and outgoing sides of a switch.
It is another object of the invention to provide a network and a switch
allowing a correct handling of particular fields identifying the type of
link, on which a data packet has arrived to a switch and on which it is to
be transmitted from the switch.
The objects mentioned above are achieved by a network and a switch
according to the invention, by means of which also the difficulties
mentioned above are avoided.
The network for the transfer of data packets thus in the common way
comprises a number of end stations, from which the data packets are issued
and which receive the data packets. The end stations comprise a connection
or terminal point having a connection line extending to an associated
switch unit. The switch units are nodes in the network and have a number
of connection points of which some can extend to end stations and other to
other switch units. Connection lines connect the different connection or
coupling points of the switch units. Each connection point is bi-directed
such that it comprises an output side and an input side. When information
is to be forwarded from an end station to another one, in some way a
connection is established in the network between these two end stations.
This connection thus is a contemplated path through the network along
which the various data packets in which the information to be transferred
is carried, are guided through the network.
All switch units are on the connection points thereof provided with ports
and in these a substitution is performed of a label in each data packet
passing through the connection line of the port. These labels are
established in the establishment of the connection and indicate the
connection to which a data packet belongs. The substitution of the label
is performed to prevent that several connections have data packets with
the same label and to simplify the handling of labels in the network, such
that each switch unit not necessarily must be provided with information in
regard of exactly all labels used in the network. The substitution of the
label of a data packet in the connection point, where the data packet
arrives to a switch unit, can be supplemented by a substitution also in
the port associated with the connection point, from which the data packet
is forwarded from the switch unit.
As has already been mentioned, the substitution of the labels is performed
to avoid that for instance data packets belonging to different
connections, in their travel through a switch unit, will be provided with
the same label. The label could in addition be used by each switch unit to
send the data packet to the correct output connection point of the switch
unit, but practically therefor a particular internal address is used being
coded to control, in the simplest possible way, the various circuits from
which the switching circuits or the switch core of the switch unit is
constructed. The substitution of the label also on an output connection
point is preferable, since hereby distributed services can be offered in a
simple way in the network. Thus for instance a data packet which arrives
to a switch unit, can be forwarded from the switch unit on several
different connection lines and on these connection lines be provided with,
when it is necessary, different labels in order that a confusion with
labels of already existent connections will not occur.
In the substitution of the label of a data packet which is performed in the
ports of the switch unit, in some way a list is established containing
records with information relating to the method of substitution of the
label of a data packet and also the new internal labels. For data packets
arriving to a port all the internal labels and the corresponding arriving
labels may thus be stored in this list, where the internal labels thus are
the labels which a data packet will carry in its travel through the switch
unit. This list can then for instance only contain all internal labels in
a suitable sequence such that an internal address in the list has a
position corresponding to the arriving label thereof, that is generally
such that the list is addressed by means of the contents of the label
field of the data packet or generally such that a table record is accessed
in some way by means of the label of the data packet. A corresponding list
may of course be arranged for the substitution of the internal labels in a
port, from which the data packets are forwarded, where thus all the
internal labels and the corresponding forwarding labels are collected in a
list.
The labels of the data packets comprise two different parts, a first part
and a second part, where only the first part is used for some connections,
say for connections of a first kind, and both the first and the second
part is used for other connections, say for connections of a second kind.
A table look-up by means of the whole label field and for the substitution
of the whole label could be performed if a sufficiently large and rapid
memory was arranged for the list used in the table look-up. In order to
reduce the requirement of memory a table look-up is performed, there is
performed a first table look-up with the first part of the label of an
arriving data packet as an entry or address in a first sublist. Hereby
information can also be obtained in regard of the kind of connection to
which the data packet belongs, i.e. if it belongs to a connection which
has only a first label part, or if it belongs to the kind of connection
which has both a first and a second label part. Further, in the first
table look-up the new label part can be obtained which is to replace the
first label part. In the case where the data packet belongs to a
connection of the second kind, it could, however, also be considered that
this substitution of the first part also is performed, but preferably the
whole label comprising the first and second part is replaced by a new or a
second table look-up in a second sublist. As an entry or address to this
second list a pointer can be used based on both information obtained in
the first table look-up and the second label part of the data packet.
Further in some way information is also inserted in the data packet in
regard of the kind of connection to Which the data packet belongs or
generally if the data packet uses only the first label field or the two
parts. This information is only retained internally in the switch unit and
is removed on the output port, from which the data packet is forwarded
from the switch unit.
The table look-up can be performed in the same way in an output port, where
the internal label is substituted by a forwarding label.
Further data packets which arrive directly from an end station or are
forwarded directly to an end station, can have a first label format, for
instance in the shape of a short field, and the data packets being
forwarded between switch units have a second label format, for instance in
the shape of a longer field, such that the short field is a part of the
longer field. In the establishment of a connection then the various label
formats are established which are valid for data packets belonging to this
connection, by the fact that indicator elements, for instance specific
data bits in some register or memory, are transferred to various signal
positions valid for the connection, i.e. if data bits are used, that they
are set or reset. When a data packet then arrives to a switch unit and to
a port thereof, it is decided in this arriving port by means of a the
appropriate indicator element, if the packet from the switch unit is to be
forwarded directly to an end station and, if it is the case, the label of
the data packet is substituted by an internal label having the first
format, and otherwise the label of the data packet is substituted by an
internal label having the second format. In the case where the first label
format is a short field being a part of a longer field, which constitutes
the second label format, when necessary for some transfers, a longer field
may be filled with data bits in the form of zeroes, such as in a transfer
from shorter to longer fields for the label.
The lists, in which the table look-up according to the above is performed,
can also be organized such that the possible contents of a data packet
arrived to or received by a port, this data packet arriving from the
outside or from the switch unit itself, are stored in a sequential order.
Then the new contents of the label field are obtained from the position or
the sequential number of the contents of the data packet in this list.
This organization of a list is chosen, when an ordinary list would require
too much memory space.
BRIEF DESCRIPTION OF THE DRAWINGS
The invention will now be described with reference to an embodiment, which
is given for explaining but not limiting the invention and which is
illustrated in the accompanying drawings, in which
FIG. 1 is a schematic picture generally showing a network,
FIGS. 2 and 3 show two different formats for the label portions of packets
in the network, where the translation thereof at the arrival to a
switching node is schematically illustrated in the lower portion of the
figures,
FIG. 4 shows the structure of a table for a part of the label which is
associated with connections established more permanently than other
connections,
FIG. 5 shows a table for the processing of a second part of the label,
FIG. 6 shows an alternative structure of a table for the processing of the
whole label of an arriving data packet,
FIG. 7 schematically illustrates the handling of a label of a data packet
in an incoming side of an interface unit of a node in the network,
FIGS. 8 and 9 show the two different formats of the label parts of packets
in the network, where the translation thereof at the forwarding from a
switching node is schematically illustrated in the lower portion of the
figures,
FIG. 10 schematically illustrates the handling of the label of a data
packet on an output side of an interface unit of a node in the network.
DETAILED DESCRIPTION
In FIG. 1 a network is schematically illustrated intended for the transfer
of data packets, such as between end stations or terminals 1. The end
stations 1 are thus both receivers and senders of the data packets, i.e.
the origins or sources and the destinations of data packets considered
here. The network in addition comprises a multitude of intermediate
switching points or switches 3. Each end station 1 has basically only one
connection point (not shown) which is bidirected and thus is arranged both
for receiving and sending data packets. This connection point or unit is
through appropriate lines coupled to a connection point 4 of a switching
unit 3. The switch units 3 have several such connection points 4, which
are also bidirected. The switching units 3 are, through some of their
connection points, connected to other similar switching or exchange units
3.
When an end station 1 requests to start sending data packets in the network
to another end station 1, a logical signal path is established through the
network. Such a path can be established by the exchange of various
messages or control information between the switch units 3. When a
connection path is established, a message can be transmitted to the end
station 1, which has requested the connection, that the connection now is
established and is ready to be used for the transfer of data packets from
the end station.
For each switch unit 3 its connection points 4 are provided by switching
ports 5, one switching port arranged for each connection point. Every
switching port 5 is divided into an input, incoming or arriving side 5a
and an output, outgoing or forwarding side 5b, such that each data packet
outgoing from a switch unit 3 will pass through the output side 5b of a
switching port 5 and a data packet arriving to a switch unit 3 will pass
through the input side 5a of a switching port 5. The physical links or
connections between a end station 1 and its associated switch unit 3, this
type of links in the figures being indicated as UNI, can be constructed
differently and have other requirements on the connection and on the
structure of the data packets than links or connections between two switch
units, which in the figures are indicated as being type NNI.
In the transport through the network the data packets are provided with a
field for identification of the packet and also containing other
information relating to the handling of the contents of the data packet,
and further the real or proper information field of the packet containing
the user information which is to be transported from the source end
station to the destination station. The identification field comprises a
label field, the general configuration of which is illustrated at the top
in FIGS. 2 and 3. The number or generally the information in this field
identifies the established connection, to which the data packet belongs.
A data packet forwarded on link from and to an end station has a label
field with the configuration as indicated at the top of FIG. 2. The label
field comprises three subfields, a field of four bits, in the figure
indicated as GFC, a field of eight bits, in the figure indicated as VPI,
and a field of sixteen bits, in the figure indicated as VCI. When a packet
comprising a label field having this configuration is forwarded from an
end station 1 to the associated switch unit, in the switching port 5 of
the switch unit, to which the data packet will first arrive in the travel
through the network, the information in the VPI field and possibly also
the information in the VCI field are replaced or translated. If the data
packet from this most adjacent switch unit 3 is to be sent back
immediately to an end station 1 without any transport through other switch
units, the field GFC of the data packet is not affected. Otherwise, when
the data packet is to be transmitted further in the network and travel
through another or several switch units 3, the GFC field is removed and
the data packet is given the configuration illustrated at the top of FIG.
3.
The actual labels are comprised or written in the fields called VPI and VCI
respectively. Some connections, called VP connections, have only
information in the VPI field, while other connections, called VC
connections, have information both in the field VPI and the field VCI.
When a data packet arrives to a switch unit 3, it may happen, that its
label as specified in the fields VPI and/or VCI coincides with the label
of another connection and that data packets of this other connection
arrive to the same switch unit 3 on another input port. To avoid a
confusion of such labels the label fields are translated to or substituted
by an internal label in the switching port or interface unit 5, on which
the data packet arrives to the switch unit 3. In the case where data
packets arrive directly from an end station, their label field thus has
the configuration as illustrated at the top in FIG. 2. The label field is
accessed in the switching port 5, i.e. in the input side 5a thereof, in
order to be processed.
At the bottom in FIG. 2 the procedure is illustrated, in this case with
data packets arriving directly from an end station, for changing the label
field in the switching port 5. From the label field first the information
carried in the VPI field is accessed. By means of this information a table
look-up in a database or register is performed. From this table look-up
information is first obtained, whether the data packet belongs to a VP
connection or a VC connection. Further also information is obtained,
whether the data packet is to be forwarded to an end station connection of
the same kind, which here as above is called a UNI link, or if the packet
is to be conveyed further in the network, the latter case as above also
being called that the packet is to be forwarded next through an NNI link.
For the case when a packet arrives on a UNI link, the packet belongs to a
VP connection and is to be forwarded directly to another end station
connected to the same switch unit, that is to a UNI link, only the
contents of the VPI field in the original data packet are substituted by a
new VPI number, which is obtained directly from the table look-up, without
any change of the length of the VPI field. If the data packet belongs to a
VP connection but instead is to be forwarded to an NNI link, the VPI field
is enlarged to comprise also the GFC field, such that the information in
the GFC field and the VPI field is replaced by a new longer VPI number
obtained in the table look-up. In the latter case, the GFC field is thus
removed and in both cases the information in the VCI field is not
affected.
If the data packet in the table look-up instead is decided to belong to a
VC connection, in the table look-up new information is obtained which is
used for an additional table look-up. At the same time other information
is also obtained, in particular as above information whether the data
packet is to be sent to a UNI link or to an NNI link. If the data packet
is to be sent to a UNI link, the information in the VPI and VCI fields is
substituted by the information obtained in this later table look-up, thus
the VPI and VCI fields of the packet being merged to one field. In the
corresponding way the GFC, VPI and VCI fields are merged to one field and
then thus the contents of these GFC, VPI and VCI fields are replaced by a
new unified VPI and VCI number, when the packet is to be forwarded to an
NNI link. In the latter case, thus also the information in the GFC field
is overwritten.
Further, both for VP and VC connections an additional information unit is
inserted in the data packet in its identification part but outside the
label field, in the shape of an addressing mode bit AM, indicating if the
data packet belongs to a VP connection or to a VC connection.
A data packet arriving to a switch unit 3 on a connection line, which is
not directly connected to an end station 1, has the configuration
illustrated at the top in FIG. 3. Here there is a VPI field comprising 12
bits and a VCI field comprising 16 bits. In the substitution or
replacement of the information in the VPI and VCI label fields inside the
identification field of the arriving data packet a procedure is performed
analogous to the one described above, with the exception that here there
is no GFC field and, that when the packet is to be sent to a UNI link, a
zeroed or reset GFC field is inserted, compare the lower part of FIG. 3.
Thus first a table look-up is performed by means of the contents of the VPI
field. From the table look-up information is first obtained if the data
packet belongs to a VP or a VC connection. In the former case also
information is obtained if the packet is to be sent to a UNI or NNI
connection. If the packet is decided to belong to a VP connection and is
to be sent to a UNI link, in the beginning of the VPI field a reset GFC
field comprising four zero bits is inserted, such that the VPI field is
shortened, and further in this new shorter VPI field information is
inserted which has been obtained in the table look-up. If the packet
instead is to be next transferred through an NNI link, the information in
the VPI field of the data packet is substituted by information obtained in
the table look-up. In these two cases the information in the VCI field is
unaffected. When it in the table look-up instead has been decided that the
data packet belongs to a VC connection, an additional table look-up is
performed based both on the information which has been obtained in the
first table look-up and by means of the information present in the VCI
field of the data packet. In this second table look-up, as above,
information is obtained, whether the data packet is to be forwarded on a
UNI or an NNI link. In the UNI case, in the beginning of the label, as
above a reset GFC field is inserted, which thus will be a part of the
original VPI field. Further a new label is obtained, which for
transmission through a UNI link has a correspondingly shorter length and
in the shape of a unified VPI and VCI number it is inserted in both the
rest of the VPI field and the whole VCI field in the original data packet.
In the transmission through an NNI link the new unified VPI and VCI number
has a whole lengt | | |