U.S. patent application number 10/252815 was filed with the patent office on 2004-03-25 for methods and devices for converting routing data from one protocol to another in a virtual private network.
Invention is credited to Chu, Thomas P., Magee, Francis R., Richman, Steven H..
Application Number | 20040059829 10/252815 |
Document ID | / |
Family ID | 31993021 |
Filed Date | 2004-03-25 |
United States Patent
Application |
20040059829 |
Kind Code |
A1 |
Chu, Thomas P. ; et
al. |
March 25, 2004 |
Methods and devices for converting routing data from one protocol
to another in a virtual private network
Abstract
Data (e.g., instructions) used to route Internet Protocol
traffic is converted from one protocol to another (e.g., MPLS to
Virtual Router) or vice-versa. This allows routers and other
network devices based on either protocol to co-exist in the same
network.
Inventors: |
Chu, Thomas P.;
(Englishtown, NJ) ; Magee, Francis R.; (Lincroft,
NJ) ; Richman, Steven H.; (Marlboro, NJ) |
Correspondence
Address: |
John E. Curtin, Esq.
Troutman Sanders LLP
Suite 600
1660 International Drive
McLean
VA
22102
US
|
Family ID: |
31993021 |
Appl. No.: |
10/252815 |
Filed: |
September 24, 2002 |
Current U.S.
Class: |
709/238 ;
709/230 |
Current CPC
Class: |
H04L 45/52 20130101;
H04L 45/00 20130101; H04L 69/08 20130101; H04L 12/4641
20130101 |
Class at
Publication: |
709/238 ;
709/230 |
International
Class: |
G06F 015/16; G06F
015/173 |
Claims
We claim:
1.) A system for converting routing data in an Internet protocol
based virtual private network (VPN), comprising: a conversion
section adapted to generate routing data in a first VPN protocol
from routing data based on a second VPN protocol.
2.) The system of claim 1, wherein the conversion section comprises
a network controller.
3.) The system of claim 1, wherein the conversion section comprises
a router.
4.) The system of claim 1, wherein the conversion section comprises
a route reflector.
5.) The system of claim 1, wherein the first VPN protocol comprises
a Multi-Protocol Label Switching (MPLS)-based protocol and the
second VPN protocol comprises a Virtual Router (VR)-based
protocol.
6.) The system of claim 5, wherein VR-based routing data comprises
a combination of data selected from the group consisting of: a PE
router Internet Protocol (IP) address; a customer site
Identification (ID); a VR ID; a VPN customer ID; a local subnet
address; a customer location subnet address; and an additional IP
subnet address.
7.) The system of claim 5, wherein the conversion section is
further adapted to generate an MPLS-based NEXT HOP parameter based
on a PE router IP address parameter.
8.) The system of claim 5, wherein the conversion section is
further adapted to generate an MPLS-based NLRI parameter based on a
customer site ID parameter.
9.) The system of claim 5, wherein the conversion section is
further adapted to generate an MPLS-based route target (RT)
parameter based on a VR ID parameter.
10.) The system of claim 1, wherein the first VPN protocol
comprises a VR-based protocol and the second VPN protocol comprises
an MPLS-based protocol.
11.) The system of claim 10, wherein MPLS-based data comprises a
combination of data selected from the group consisting of: a NEXT
HOP parameter; an NLRI field, an RT; and a route distinguisher.
12.) The system of claim 10, wherein the conversion section is
further adapted to generate a VR-based PE router IP address
parameter based on a NEXT HOP parameter.
13.) The system of claim 10, wherein the conversion section is
further adapted to generate a VR-based customer site ID parameter
based on a NLRI field.
14.) The system of claim 10, wherein the conversion section is
further adapted to generate a VR-based ID parameter based on a
route target parameter.
15.) The system of claim 10, wherein the conversion section is
further adapted to generate a VR-based VPN customer ID parameter
based on an RT parameter.
16.) A method for converting routing data in an Internet protocol
based virtual private VPN, comprising: generating routing data in a
first VPN protocol from routing data based on a second VPN
protocol.
17.) The method of claim 16, wherein the first VPN protocol
comprises a MPLS-based protocol and the second VPN-based protocol
comprises a VR-based protocol.
18.) The method of claim 17, wherein VR-based routing data
comprises a combination of data selected from the group consisting
of: a PE router Internet Protocol (IP) address; a customer site
Identification (ID); a VPN customer ID; a VR ID; a customer
location subnet address; and an additional IP subnet address.
19.) The method of claim 17, further comprising generating an
MPLS-based NEXT HOP parameter based on a PE router IP address
parameter.
20.) The method of claim 17 further comprising generating an
MPLS-based NLRI parameter based on a customer site ID
parameter.
21.) The method of claim 17, further comprising generating an
MPLS-based RT parameter based on a VR ID parameter.
22.) The method of claim 16, wherein the first VPN protocol
comprises a VR-based protocol and the second VPN protocol comprises
an MPLS-based protocol.
23.) The method of claim 22, wherein MPLS-based data comprises a
combination of data selected from the group consisting of: a NEXT
HOP parameter; an NLRI field, an RT; and a route distinguisher.
24.) The method of claim 22, further comprising generating a
VR-based PE router IP address parameter based on a NEXT HOP
parameter.
25.) The method of claim 22, 4 further comprising generating a
VR-based customer site ID parameter based on a NLRI field.
26.) The method of claim 22, further comprising generating a
VR-based ID parameter based on a route target parameter.
27.) The method of claim 22 further comprising generating a
VR-based customer ID parameter based on an RT.
Description
BACKGROUND OF THE INVENTION
[0001] Historically, when a company or any other group of people
needed to have their computers and the like "networked" together
they would contact a local telephone company or another so-called
"service provider" ("SP") to assist them. The service provider
would design and construct a network using so-called
"connection-oriented" technology (e.g., some combination of leased
"private lines" and the publicly switched telephone network,
dial-up lines or the like).
[0002] As is known by those skilled in the art, networks designed
using existing connection-oriented technologies have their
drawbacks.
[0003] In order to overcome the drawbacks inherent in existing
network designs, SP based, Internet-protocol ("IP") virtual private
networks ("VPNs") have been developed ("IP-VPNs" for short).
IP-VPNs are designed using "connectionless" technology. One of the
advantages a connectionless network has over a connection-oriented
network is that there is no need for an administrator or the like
of a network to specify traffic characteristics between two sites
or locations (hereafter collectively referred to as "location(s)")
in a network. Instead, it is now up to the SP to deliver
communication services that are associated with a certain "Quality
of Service" ("QoS") level.
[0004] This allows an SP to manage all of the traffic flowing from
each of its customers as an aggregate, resulting in increased
efficiencies in both network resource usage and network
management.
[0005] Two common techniques used to create an SP based, IP-VPN are
Multi-Protocol Label Switching ("MPLS") and "virtual routers"
("VR"). It should be understood that both techniques are used to
implement VPNs. The MPLS approach is articulated in an Internet
protocol proposal Request for Comment ("RFC") 2547 (RFC 2547)
entitled "BGP/MPLS VPN'S" (as well as in internet draft-rfc2547bis,
its second version). The VR approach is in actuality, a family of
techniques. There are a number of possible ways to implement a
VR-based, VPN. One implementation is articulated in Internet RFC
2917 entitled "A Core MPLS IP VPN Architecture".
[0006] Overly simplified, the difference between an MPLS-VPN and a
VR-VPN is that the former uses so-called route distinguishers (RDs)
and route targets (RTs) to route communications traffic (e.g.,
data) from one location in a network to another, while the latter
uses so-called "access lists" to accomplish the same thing.
[0007] Before discussing the details of the present invention, it
may be helpful to introduce some terms which will be used
repeatedly throughout the discussion below.
[0008] In RFC 2547 (i.e., MPLS-based techniques), an IP network is
divided into two tiers, a core network that consists of "core
routers", and edge networks that consist of "service provider edge"
("PE") routers. Customer routers are in turn connected to the PE
routers. The customer routers that are directly attached to the PE
routers are referred to as "customer edge" ("CE") routers. All VPN
functions are implemented in the PE routers. Core routers are
operable to forward MPLS "packets", (e.g., small bits of data) but
they are not assigned VPN tasks. Similarly, CE routers behave as if
they are connected to ordinary routers, (e.g., they do not receive
information telling them that PE routers are RFC 2547
compliant).
[0009] In RFC 2547, a customer location is connected to a PE router
through a CE router and the connection is identified via a layer 1
or a layer 2 identifier that can represent: a physical interface
ID; a virtual path/virtual circuit identifier of an ATM interface
("ATM" stands for Asynchronous Transfer Mode); a data link
connection identifier of a frame relay interface; a virtual local
area network identifier of an Ethernet serial link interface;
and/or the MPLS label of a MPLS interface. One or more of these
interfaces will be referred to hereafter as a "pathway".
[0010] A basic requirement for a VPN is that each IP-VPN subscriber
must be able to use its own private IP addressing scheme.
Therefore, each PE router needs to be able to route IP packets
based on differing incoming data streams. In theory, this may
require a different decision process for each data stream. There
are two possible approaches which can be used by a PE router. The
first is to create a "routing/forwarding table" for each VPN. The
second is to create a single routing/forwarding table with
"context" for each VPN. In RFC 2547, the first approach is
considered resource and management intensive, so the second
approach is utilized.
[0011] Routing tables are stored within each PE router. It is the
routing tables that contain the instructions, guidelines and the
like which tells the PE router how to treat each received data
stream. That is to say, each routing table provides directions, for
example, on how to handle an incoming data stream, where to route
it next, if any action should be taken at all, etc. . . .
[0012] The context specific table for each VPN is referred to as a
VPN Routing and Forwarding ("VRF") table. Each VRF table is
identified by a parameter known as a Route Distinguisher ("RD").
For the sake of efficiency, multiple data streams from different
pathways can point to the same VRF. An RD contains two fields that
identify the SP and the routing domain within the SP's network. RD
assignment is the responsibility of the SP.
[0013] To create such VRF tables manually is cost prohibitive and
not scaleable. Therefore, a "routing protocol" between PE routers
is used to automatically update and synchronize the content of the
VRF tables each time locations in the network are added, deleted or
modified. In RFC 2547, the routing protocol used is a Border
Gateway Protocol with Multi-Protocol Extensions ("BGP-MP"), as
specified in RFC 2858 from the Internet Engineering Task Force.
[0014] The BGP-MP routing protocol specifies a number of
parameters, one of which is called a Route Target ("RT"). A PE uses
RTs to "advertise" its routes to other PEs that are considered its
"peers". RTs are used to describe the VPN (or "VPN" component) that
the route is applicable to. Because a location may belong to
multiple VPNs or VPN components, multiple RTs can be associated
with a single route.
[0015] Both RDs and RTs are known as network wide parameters
because they have to be unique across the entire network. To ensure
that a VPN is working properly, RDs and RTs must be properly
generated and assigned. If the incorrect RT is assigned it may be
impossible for one or more PE routers to correctly route packets of
data. Furthermore, if RDs are efficiently assigned the number of
VRFs in a PE can be reduced, thereby allowing network resources to
be conserved. Further still, with efficient RT assignment, it would
be unnecessary to reconfigure existing PE routers each time a new
location is added resulting in a considerable savings in network
management.
[0016] RD and RT assignments will change over time because: (1) the
networks they are associated with will change as new locations are
added, deleted or modified within such networks; and/or (2) the
rules governing the flow of data to and from such locations will
change.
[0017] The proper assignment of RDs and RTs is addressed in
companion U.S. patent application Ser. No. ______, the disclosure
of which is mostly repeated herein or incorporated herein by
reference.
[0018] In the discussion just concluded we focused on MPLS-based
techniques for routing data from one VPN to another. As stated
earlier, there exists a second technique, the VR approach. We now
turn our attention to that technique.
[0019] The VR technique involves the generation and use of a number
of logical routers (i.e., software, firmware configured to carry
out the functions and features of one or more physical routers . .
. this technique is sometimes called "emulating" a physical router
. . . ). Each logical router is adapted to exhibit the behavioral
characteristics of separate physical routers. These logical routers
are aptly referred to as virtual routers ("VRs"). In an IP-based
VPN, each VPN is assigned a VR within each PE router. The VRs can
be connected to each other via a core network and a number of layer
2/layer 3 technologies, such as ATM virtual connections, frame
relay connections, IP encapsulation and Layer 2 Tunnel
Protocol.
[0020] Sometimes an SP's customer places restrictions on locations
in its network (e.g., not all locations are allowed to communicate
with each other directly). Because of this, a VR must be able to
"filter" packets. That is to say, a VR must be able to, for
example, determine whether to forward or discard a packet. The
filtering capabilities of VRs are specific to the configuration of
each VPN. Most VRs filter packets using at least the fields
containing the IP source address, IP destination address, source
and destination ports, protocol type and the type of service
("TOS") byte embedded within each packet.
[0021] More specifically, each VR is linked or otherwise has access
to, one or more "access list(s)". These lists contain the exact
instructions on how a VR should treat a received packet.
[0022] An access list usually comprises a number of entries or
"statements", each of which defines whether a particular packet
will be forwarded or dropped based on whether the particular packet
satisfies certain criteria. For example, an access list can specify
that only packets with particular destination addresses are
forwarded, while all others are discarded.
[0023] Access lists are assigned to a particular pathway (i.e., a
particular interface which connects a PE and CE router) in a
particular order. In general, for an incoming packet a VR is
adapted to match the data packet with criteria specified in a first
entry of a first access list. If a match is found, an action
specified in the entry will be executed. If a match is not found,
the VR is adapted to proceed to the next entry in the access list
until a match is found or until the end of the list is reached.
[0024] There are two access lists, one transmission and one
reception, associated with each pathway. These access lists are
stored in PE routers. These two access lists are referred to as
"master access lists" or "master lists". Each master list, in turn,
comprises a number of sub-access lists. When a new VPN or VPN
component is added or deleted, a sub-access list has to be added or
deleted from a master access list. In addition, when changes are
made to an existing VPN or VPN component (e.g., modifying or
removing a location), entries in the corresponding sub-access
access lists have to be changed manually. This is labor intensive
and leads to many mistakes.
[0025] The automatic generation and assignment of correct VR access
lists is also addressed in companion U.S. patent application Ser.
No. ______ much of which is also repeated herein or incorporated
herein by reference.
[0026] In an actual IP-VPN it is possible, and even likely that
within an SP's network both MPLS-based and VR-based routers exist.
It is also likely that a customer may desire for an SP to deliver
communication services between MPLS and VR-based routers. However,
current techniques are cumbersome and, in many ways, inefficient
for doing so. Therefore, it is desirable to provide techniques for
providing interoperability between VR and MPLS-based routers.
[0027] Further it is desirable to provide techniques for converting
data in MPLS-based protocols to data-based on VR-based protocols
and vice-versa.
BRIEF DESCRIPTION OF THE DRAWINGS
[0028] FIG. 1 depicts a simplified block diagram of a network used
to illustrate features of the present invention.
[0029] FIG. 2 depicts three basic types of components making up a
VPN.
[0030] FIG. 3 depicts a simplified flow diagram of a technique for
identifying full mesh components of a VPN according to one
embodiment of the present invention.
[0031] FIG. 4 depicts an example of a network of VPNs.
[0032] FIG. 5 depicts an example of a connectivity matrix generated
according to one embodiment of the present invention.
[0033] FIG. 6 depicts a simplified, iterative flow diagram of a
technique to identify root-receiver and root-transmitter components
of a VPN according to one embodiment of the present invention.
[0034] FIG. 7 depicts an example of a second connectivity matrix
generated according to one embodiment of the present invention.
[0035] FIG. 8 depicts a simplified flow diagram of a technique for
identifying basic components of an MPLS-based VPN and the
generation and assignment of RDs and RTs according embodiments of
the present invention.
[0036] FIG. 9A depicts a simplified block diagram of a device for
converting routing data from MPLS-based data to VR-based data and
vice-versa according to one embodiment of the present
invention.
[0037] FIG. 9B depicts another simplified block diagram of a device
for converting routing data from MPLS-based data to VR-based data
and vice-versa according to another embodiment of the present
invention.
[0038] FIG. 10 depicts a simplified table illustrating the
conversion of VR-based routing data to MPLS-based routing data
according to embodiments of the present invention.
[0039] FIG. 11 depicts a simplified table illustrating the
conversion of MPLS-based routing data to VR-based routing data
according to embodiments of the present invention.
SUMMARY OF THE INVENTION
[0040] In accordance with the present invention, a conversion
section is adapted to generate routing data (e.g., instructions) in
a first protocol (MPLS or VR) from routing data based on a second
protocol (VR or MPLS).
DETAILED DESCRIPTION OF THE INVENTION
[0041] As envisioned by the present inventors, one embodiment of
the present invention comprises a network controller which is
adapted to first identify basic components of a VPN and then to
generate routing data for each of the identified components.
[0042] In addition, the present invention envisions one or more
routers, such as PE routers, each of which is adapted to route
communication traffic based on the generated routing data received
from the network controller.
[0043] The present invention envisions a number of ways to identify
basic components of a VPN. To simplify the explanation which
follows, a technique such as the one described in U.S. patent
application Ser. No. ______ may be used. It should be understood,
however, that other techniques may be used as well.
[0044] The discussion which follows will first concentrate on the
identification of these basic components. Thereafter, two
techniques for providing interoperability (e.g., by converting
routing data) will be discussed. The first technique being an MPLS
to VR-based technique; the second being a VR to MPLS-based
technique.
[0045] Referring now to FIG. 1, there is shown a simplified block
diagram of a network 1. As shown, the network 1 comprises a
controller 100 and one or more PE routers 110a, 110b . . .
110.sub.n (where ".sub.n" represents the last PE router making up a
network).
[0046] In one embodiment of the present invention, the network
controller 100 comprises an identification section 101 and a data
routing generator 102. The identification section 101 is adapted to
identify the basic components of all of the VPNs which are a part
of network 1. Routing generator 102 is adapted to generate routing
data (e.g., routing instructions) for each of the components
identified by section 101.
[0047] Once the routing generator 102 has generated the appropriate
data, it is sent to the appropriate PE router(s) 110.sub.a,
110.sub.b . . . 110.sub.n. In additional embodiment of the present
invention, the controller 100 comprises an interface section 103
which is adapted to transfer the generated data to one or more of
the PE routers 110.sub.a, 110.sub.b, . . . 110.sub.n.
[0048] It should be understood that the basic components identified
by the identification section 101 can take on a variety of
different configurations. Typically, an SP (e.g., telephone
company) which is operating network 1 must obtain a description of
a customer's VPN. Depending on the customer, some may or may not be
capable of providing the SP with the description of their VPNs or
the basic components making up their VPNs. In fact, typically a
customer is only capable of describing their network in terms of
"connectivity" requirements between various locations in the
network. For this reason, the present inventors have envisioned a
technique which will allow the identification section 101 to
identify these basic components. Again, as stated before, this
technique is only one of many which may be used.
[0049] Before turning to the technique itself, it may be useful to
discuss, generally speaking, the different types of components
which typically make up a VPN. The first component is a "full mesh"
component. It is the most common and basic type of component in a
VPN. A full mesh component can be generated (i.e., created) by
grouping together locations belonging to a particular VPN component
that are permitted to send and receive packets to, and from, one
another.
[0050] There are two other types of basic components. One is called
a "root-receiver" component (root-receiver for short) and the other
is called a "root-transmitter" component (root-transmitter for
short). When many locations are logically connected to one common
location, the common location is called a "root". Both
root-transmitter and root-receiver components have a tree-like
topology where a location acts as the "root" of the tree and a
number of locations act as the "leaves" of the tree. In a
root-receiver VPN component, all of the leaves are allowed to
transmit to a root. However, a root cannot transmit to the
leaves.
[0051] A root-transmitter is substantially the opposite of a
root-receiver. That is, a root-transmitter is identified by the
fact that the root is allowed to transmit to a number of leaf
locations but the leaves cannot transmit to the root. Each of the
three basic types of components are shown in FIG. 2.
[0052] Referring now to FIG. 3, there is shown a simplified flow
diagram of one technique for identifying all full mesh components
of VPNs in a given network, some of which will be discussed
below.
[0053] In one embodiment of the present invention, the
identification section 101 is adapted to generate a first
connectivity matrix based on two-way communication paths associated
with some or all of the locations in one or more VPNs. One example
of a network 1000 of VPNs is shown in FIG. 4 while its associated
connectivity matrix is shown in FIG. 5. In one embodiment of the
present invention, each row and column in the connectivity matrix
shown in FIG. 5 represents a location in network 1000. As will be
recognized by those skilled in the art, this first connectivity
matrix is symmetrical because each connection is a bi-directional
connection. In addition, any location in the network 1000 can send
to itself thereby making the main diagonal of the matrix equal to
"1". It should be understood that the terms "bi-directional
connection" and "full mesh component" are used to describe the same
set of circumstances in a given network. In an illustrative
embodiment of the present invention, section 101 is adapted to
construct the first connectivity matrix so that the identification
section 101 can identify all maximal, full mesh components
associated with VPNs within network 1000 from the first matrix.
[0054] Before going further, it should be noted that although the
identification section 101, routing generator 102 and interface
section 103 are shown as separate sections within controller 100,
it should be understood that these sections may be combined into
fewer sections or further broken down into additional sections.
Also, each "section" may, in fact, comprise one or more software or
firmware programs adapted to carry out the features and functions
of the present invention. That is not to say that the sections
making up the controller 100 shown in FIG. 1 must be realized
completely in software or firmware. Rather, the sections shown in
FIG. 1 may be realized in a combination of hardware, software,
firmware or the like.
[0055] Full mesh components are only one type of basic component.
In one embodiment of the invention, after the identification unit
101 has identified all of the full mesh components it is further
adapted to identify all root-receiver and root-transmitter
components. Root-receiver and root-transmitter components are
extracted from one-way connections associated with the VPNs shown
in FIG. 4 (not two-way, as is the case for full mesh components).
FIG. 6 depicts a simplified, iterative flow diagram of a technique
which may be used to identify root-receiver and root-transmitter
components according to one embodiment of the present
invention.
[0056] Initially, the identification section 101 is adapted to
identify all one-way transmission connections from the
VPNs/locations shown in FIG. 4 and then to generate a second
connectivity matrix. An example of such a matrix is shown in FIG.
7. The root-receiver and root-transmitter VPNs are extracted from
the one-way connections represented by this second matrix according
to the rules which follow.
[0057] As noted above earlier, the technique just outlined for
identifying all of the basic components of the VPNs within a
network is one of many that can be used. In an alternative
embodiment of the present invention, the identification section 101
is adapted to identify the root-transmitter components first and
then the root-receiver components. Further, section 101 can be
adapted to identify any full mesh component that has only two
members ( . . . identified using the first matrix . . . ) which
share a common location and to convert or combine such components
into a pair of one way connections (i.e., a root-receiver and/or a
root transmitter component) according to a set of rules:
[0058] a. First, identify all the two-member full-mesh VPN
components that share a common member (or location);
[0059] b. convert the above group into a single root-transmitter
and a root receiver VPN component with the common member (or
location) as the root; and
[0060] c. repeat the two steps immediately above, until all
two-member, full-mesh VPN components having common members are
identified.
[0061] As will be apparent to those skilled in the art, the initial
steps in the technique described above amount to a systematic
method for identifying the basic components of a network. Deciding
which technique is the "best" technique, depends on many factors,
such as the evolution of the network. It should be understood that,
the greater the detail and accuracy of the information supplied to
an SP by a customer concerning its VPN(s), the better the results
will be using the identification technique described above.
[0062] After the basic components of a network have been
identified, the present invention envisions that the data routing
generator 102 shown in FIG. 1 is adapted to generate routing data
based on the components identified and their relationships to one
another. Routing generator 102 is adaptable to generate routing
data for both MPLS and VR based-VPNs.
[0063] We will discuss MPLS-based VPNs first. FIG. 8 depicts a
simplified flow diagram of a technique for identifying the basic
components of an MPLS-based VPN as well as the generation and
assignment of RDs and RTs for an MPLS-based VPN according to one
embodiment of the invention.
[0064] In one embodiment of the present invention, the routing
generator 102 shown in FIG. 1 comprises an RD assignment section
104 adapted to generate and assign RDs to network components
identified by the identification section 101 and an RT assignment
section 105 adapted to generate and assign RTs.
[0065] An RD is a value or a "tag" so to speak that is assigned to
each PE router 110.sub.a, 110b . . . 110.sub.n, by the controller
100 via section 103. Upon receiving an RD and RTs, a router
110.sub.a, 110.sub.b . . . 110.sub.n, is adapted to generate VRF.
Generally, an RD value contains two fields: a field that identifies
the SP and a field that identifies the RD in the SP's network. In
one embodiment of the invention, the RD assignment section 104 is
adapted to assign one RD to all locations that are members of the
same, identified full-mesh VPN component. In addition, the section
104 is adapted to assign a single, unique RD to each location which
is part of a root-receiver or root transmitter, identified
component.
[0066] As mentioned above, an RT assignment section 105 is adapted
to assign RTs. In one embodiment of the present invention, section
105 is adapted to assign RTs as follows:
[0067] Each full mesh VPN component is assigned its own RT. This
allows each member of a full mesh VPN component to "export" and
"import" the assigned RT;
[0068] Each root transmitter VPN component is assigned its own RT.
This allows each branch of a VPN component to export the RT, and
the root of the VPN component to import the RT; and
[0069] Each root receiver VPN component is assigned its own RT.
This allows each branch of a VPN component with to import the RT,
and the root of the VPN to export the RT.
[0070] The terms "import" and "export" are described in RFC 2547.
In general, if a location wants to receive packets associated with
a particular VPN component, it would export the RT associated with
the particular VPN by "advertising" the RT via a routing protocol.
If a location is authorized to transmit data associated with a
particular VPN component, it would import a route advertisement,
with a matching RT, into its VRF table.
[0071] Backtracking somewhat, in another embodiment of the
invention, the identification section 101 can be adapted to
identify all root-transmitter and root-receiver VPN components that
have the same branches. Two root-transmitter (or two root-receiver)
VPN components that have the same branches are referred to as being
"equivalent". It can be said that the section 101 can be further
adapted to associate components having the same branches to an
equivalent root-transmitter or root-receiver component (which ever
applies).
[0072] Consistent with such an embodiment the route generator 102
can be adapted to assign the same RT to an equivalent root
transmitter (or root receiver) VPN component and to assign the same
RD to the roots of equivalent root transmitter (or root-receiver)
VPN components so as long as all of the roots belong to the same
full mesh VPN component.
[0073] In one embodiment of the invention, the network interface
section 103 shown in FIG. 1 is adapted to transfer the assigned RDs
and RTs (i.e., routing data) to the one or more routers 110.sub.a,
110.sub.b . . . 110.sub.n, using a Simple Network Management
Protocol ("SNMP") or similar protocol. The routers 110.sub.a,
110.sub.b . . . 110.sub.n, are then adapted to construct VRF tables
based on the transferred RDs, RTs and BGP-MP route advertisements
as specified in RFC 2547. Once the VRF tables are constructed, the
routers 110.sub.a, 110.sub.b . . . 110.sub.n are adapted to route
communications traffic using the so-constructed tables.
[0074] As the configuration of network 1000 changes and new basic
components are formed, the RD and RT assignments must, for the most
part, change as well. In one embodiment of the invention, the
controller 100, via section 103 is adapted to repeatedly transfer
updated routing data (e.g., RDs and RTs) to the routers 110.sub.a,
110.sub.b . . . 110.sub.n
[0075] The routing generator 102 is also adapted to generate
routing instructions for VR-based VPNs. In a VR-based IP-VPN, each
VPN is assigned a single VR within a PE router. It understood that
each PE router may comprise a number of VRs, one for each VPN. It
should be further understood that if a customer of an SP has more
than one VPN, each VPN is assigned its own VR.
[0076] VR-based VPNs do not use RDs and RTs. Instead, "access
lists" are used to control the flow of traffic. These access lists
are generated by the generator 102 and transferred to routers
110.sub.a, 110.sub.b . . . 110.sub.n, via section 103 so that each
VR in a PE router can control the routing of communications
traffic.
[0077] As mentioned above earlier, associated with each pathway
between a customer location and a VR are two access lists, one for
transmitting traffic to, and one for receiving traffic from, a
network.
[0078] In one embodiment of the invention, the generator 102 is
adapted to generate and transfer to each router 110.sub.a,
110.sub.b . . . 110.sub.n, two master access lists for each such
pathway (so-called "RECEIVE" and "SEND" lists discussed below).
Further, each of the master access lists comprises a number of
sub-access lists each sub-access list being associated with a
single VPN component.
[0079] In one embodiment of the invention, when an incoming packet
arrives at a router, the router is adapted to forward the packet to
a first VR (i.e., within the PE). Thereafter, the VR is adapted to
compare the incoming packet with criteria specified in the first
entry of the first access list. If there is a match, the router is
adapted to execute or otherwise carry out the actions as specified
in the entry. If not, the VR is adapted to proceed to the next
entry in the first access list until a match is found or the last
access list is exhausted. Thereafter, the incoming packet is
discarded and the next incoming packet undergoes a similar
comparison process.
[0080] The operation of the routing generator 102 in a VR-based VPN
mode will now be discussed in more detail. According to one
embodiment of the present invention, the routing generator 102 is
adapted to generate a RECEIVE and a SEND master access list of
Internet protocol addresses based on VPN components identified by
the identification section 101 for each network pathway. The
routing generator 102 is further adapted to reduce the size of the
access lists by substantially summarizing the Internet addresses.
Address summarization involves combining two or more IP address
"subnets" into a single larger IP address subnet. In addition, the
generator 102 is adapted to generate revised, master access lists
each time a location (or VPN component) is modified, added to, or
deleted from, a VPN (or network).
[0081] To generate the master access lists, the generator 102 is
further adapted to generate secondary RECEIVE and SEND access
lists, one for each VPN component identified by the identification
section 101. It should be understood that although the master
access lists have been discussed first, in reality the generator
102 is adapted to generate the secondary access lists first and
then combined these lists to form the master access lists. The VPN
components may be full-mesh, root-receiver or root-transmitter
components. As with the master access lists described above, the
generator 102 is further adapted to generate revised secondary
lists each time a location (or VPN component) is modified, added
to, or deleted from a VPN component (or network).
[0082] It should be understood that, after the generator 102 has
generated the lists, the network interface section 103 is adapted
to transfer the master/secondary RECEIVE and SEND lists to one or
more routers 110.sub.a, 110.sub.b . . . 110.sub.n, in the network.
The routers are then adapted to route communication traffic based
on the transferred RECEIVE and SEND access lists.
[0083] It should be noted that different customer locations
connected to a VR may have the same access list. In particular,
locations that, in RFC 2547 terminology, export and import the same
RTs, would have the same access lists. Taking this into account, it
is possible to generate a reduced number of access lists.
[0084] Once routing data is generated it is passed or routed from
one router to another. Because many networks may have, or may
desire to have, both MPLS and VR-based routers, there is a need to
ensure that these routers can route data between them effectively.
FIG. 9A is a simplified block diagram of a device 900 for
converting routing data from a first protocol to a second protocol
and vice-versa (e.g., from MPLS-based data to VR-based data)
according to embodiments of the present invention (it being
understood that "first" may refer to either protocol and "second"
may refer to either protocol). In one embodiment, the system 900
comprises conversion section 902 adapted to generate new routing
data comprising a next set of parameters based on parameters
contained within routing data of another protocol. In one
embodiment, the new routing data is MPLS-based routing data; in
another embodiment the new routing data is VR-based routing
data.
[0085] The conversion section 902 may comprise one or more
platforms (e.g., hardware, software, firmware or any combination of
these elements) adapted to implement the various features and
functions of the present invention described above and below.
[0086] In another embodiment of the present invention, the
conversion section 902 comprises a "route reflector". A route
reflector comprises a logical module, or plurality of modules, that
is connected to a VPN and is adapted to serve as a concentration
point for edge routers. In one embodiment, the route reflector may
be implemented in a designated router. A route reflector is
operable to receive messages from client components that it serves
and is adapted to forward the messages throughout the VPN.
Conversely, the route reflector is further adapated to receive
messages from the rest of the VPN and forward them to the
appropriate client component.
[0087] FIG. 9B is a simplified block diagram illustrative of yet
another embodiment of the present invention wherein, the system 900
may comprise an identification section, like section 101, adapted
to identify basic components of the network as described above.
After the identification section 101 has identified the basic
components of the network and the conversion section 902 has
generated the appropriate routing data, a router interface section
904 can be adapted to transmit the generated routing data to at
least one of the identified VPN components.
[0088] We now turn to a more detailed description of the operation
of the conversion section 902. First, we will describe the
operation of a system where VR-based routing data is converted to
MPLS-based routing data. Then, we will describe the operation of a
system where MPLS-based routing data is converted to VR-based
routing data.
[0089] In one embodiment of the present invention, VR routing data
comprises a combination of data selected from a PE router IP
address, a customer site ID, a VR ID, a customer location subnet
address and an additional IP subnet address. FIG. 10 depicts a
simplified diagram illustrating a technique for generating MPLS
routing data from VR routing data. As shown, the conversion section
902 is adapted to generate an MPLS-based NEXT HOP parameter based
on a PE router IP address parameter, an MPLS label parameter based
on a customer site ID parameter, an MPLS-based NLRI parameter based
on a customer location IP subnet address parameter, and an
MPLS-based RT parameter based on a VR ID parameter. The RD
parameter maybe input manually by the user for each customer
site.
[0090] In an alternative embodiment of the present invention,
MPLS-based routing data comprises a combination of data selected
from a NEXT HOP parameter, an NLRI field, an RT and an RD.
[0091] FIG. 11 depicts a simplified diagram illustrating a
technique for generating VR-based routing data from MPLS-based
data. In an embodiment of the present invention, the conversion
section 902 is yet further adapted to generate a VR protocol PE
router IP address parameter based on an MPLS-based NEXT HOP
parameter, a VR protocol customer site ID parameter based on an
MPLS label, a VR ID parameter based on an RT lookup table, a VPN ID
based on an RT lookup table and a customer location local IP subnet
address based on an NLRI field. Additional IP subnets address are
obtained from VPN-specific processing. In an alternative
embodiment, the VR ID parameter and the VPN customer ID may be
obtained by decomposing an RT field into VR and VPN customer
subfields rather than by accessing a lookup table.
[0092] The examples given above have sought to illustrate the
conversion of routing data from MPLS-based data to VR-based data
and vice-versa. Alternate embodiments will become apparent to those
skilled in the art to which the present invention pertains without
departing from its spirit and scope as defined by the claims which
follow:
* * * * *