| The MPLS forwarding procedure is very similar to the "label
swapping" procedure in ATM switches. They use the input port and
the incoming VPI/VCI value as an index into a "cross-connect"
table, from which they obtain an output port and an outgoing VPI/VCI
value. Therefore, if MPLS labels can be encoded into the fields which
are accessed by these ATM switches, they could be used as LSRs.
We call these devices as "ATM-LSRs". |
| |
|
MPLS labels could be encoded into AAL5 ATM switches as
follows: |
| |
- SVC encoding: The VPI/VCI field is used to encode the
top MPLS label. This technique can be used in any network. Each
LSP is realized as an ATM SVC and the LDP becomes the
ATM "signalling" protocol. However, the ATM switch
cannot perform "push" and "pop" operations on the label
stack. Only one label can be encoded using this method. Label stack is not
applicable.
- SVP encoding: The VPI field is used to encode the top
label and the VCI field is used to encode the second label, if one
is present. This technique permits the use of ATM "VP-switching"
where LSPs are realized as ATM SVPs, with the LDP
serving as the ATM signaling protocol. However, when the network
includes an ATM Virtual Path through a non-MPLS ATM network,
the VPI is not necessarily available to MPLS and the
technique can not be used. When using this technique, the ATM-LSR
at the egress of the VP effectively does a "pop" operation.
- SVP Multipoint encoding: The VPI field is used to encode
the top label, and part of the VCI field is used to encode the
second label, is one is present, and use the remainder of the VCI
field to identify the LSP ingress. Then conventional ATM
VP-switching capabilities can be used to provide
multipoint-to-point VPs, which enables label merging. If there are
more labels on the stack than can be encoded in the ATM header, the
encoding must be combined with generic encapsulation.
|
|
Interoperability among Encoding Techniques |
| |
| Having an LSP's segment <R1,R2,R3>, it is possible that R1
uses a different encoding label technique to send a packet to R2,
that the encoding technique used by R2 to send a packet to
R3. Then, MPLS architecture supports LSPs with different
label stack encoding techniques. The labeled packet processing
is abstract for us. When a labeled packet is received, the
LSR must decode it to determine the current label value, then using this
value will determine a new value for the stack, and finally will encode this
new value appropriately before transmitting the labeled packet to its
next hop. |
|
|
|
|
|
ATM switches have no capability for translating from one
encoding technique to another. Because of this, two ATM switches
being sucessive LSRs along a level m LSP, must use for this
level the same encoding technique. When combining in one LSP,
LSR's ATM switches with LSR's shim header switches, some of
these switches may swap off an ATM encoded label stack on an
incoming interface, and replace it with an MPLS shim header encoded label
stack on the outgoing interface. These LSR switches are called "hibrid"
switches. |
|
