LDP (Label Distribute Protocol): The label distribution protocol allocates labels based on destination addresses and uses label forwarding instead of IP forwarding. The advantage of LDP:It isolates public network routes, selects forwarding paths based on optimal routes, and supports ECMP. And the configuration is simple.
Different types of services, such as voice, data, video, and VR, have different requirements on the network. The bandwidth requirements are increasing, and the network scale is increasing explosively. The traditional MPLS technology requires a dedicated label distribution protocol. Labels need to be allocated to each LSP, which occupies a large amount of resources. Status maintenance protocol packets occupy a large amount of bandwidth. It needs to be synchronized with the IGP protocol. The deployment and maintenance are complex and the scalability is poor. This network operation mode cannot meet the requirements of service providers to rapidly deploy network services on demand. In addition, the OPEX (Operating Expense) increases linearly with scale.
LDP depends on the IGP routing table to calculate LSPs,
RSVP:Resource Reservation Protocol. RSVP-TE is introduced to solve the problem that the traditional IP network can forward only the optimal path and the path cannot be planned. RSVP-TE brings many benefits, such as explicit path planning, bandwidth resource reservation, and multiple protection schemes.
- RSVP relies on IGP to maintain its neighbor and link status, complicates the control plane, and complicates network maintenance and fault location.
- RSVP-TE needs to establish multiple tunnels to implement the ECMP function.
- In the LSP status maintenance, PATH and RESV need to be updated continuously.
After services increase, performance problems occur at the intermediate point (services are overlapped.In addition, the configuration quantity of the RSVP-TE tunnel is always criticized by the operator. An average of eight tunnels need to be configured for each tunnel.
C:Control
F:Forwarding
SDN (Software Defined Network) is a new network innovation architecture proposed by the Stanford University clean slate Research Group. The core technology separates the control plane from the data plane to implement flexible control of network traffic and provide a good platform for the innovation of core networks and applications.
The revolutionary SDN network uses a centralized control architecture. The control functions (such as route calculation) of network devices are centralized on one controller, and the forwarding table is generated and delivered by the controller to the device. Device functions are simplified and are only responsible for forwarding. OpenFlow is a control interface between the Controller and devices.
The incremental SDN network expands the existing network. The existing devices evolve to the SDN. Some control functions are reserved on the device. Some functions that require centralized control are processed on the controller, which emphasizes the smooth evolution capability of the device.
Prefix Segment indicates the destination address, and Adjacency Segment indicates the outgoing link of the data packet, which can be similar to the destination IP address and outbound interface in the traditional IP forwarding. In an IGP area, the network element device floods the Node SID and the Adjacency SID of the network element device by using an extended IGP message, so that any network element can obtain information of another network element.Any path on the network can be constructed by combining the prefix (node) SID and adjacent SID in sequence. In each hop of the path, the next hop is distinguished by using the stack top segment information. The segment information is stacked on the top of the data head in sequence. When the stack top segment information includes the identifier of another node, the receiving node uses the equivalent multipath (ECMP) to forward the data packet to the next hop. When the stack top segment information is the identifier of the current node, the receiving node pops up the top segment and executes the task required by the next segment.
In actual applications, Adjacency Segment, Prefix Segment, and Node Segment can be used independently or together.
There are three scenarios: Prefix Segment, Adjacency Segment, Adjacency Segment + Node Segment
Segment Routing encapsulates the segment sequence representing the forwarding path into a data packet header and transmits the packet with the data packet. After receiving the data packet, the network node parses the segment sequence. If the top segment identifier of the segment sequence is Node SID, the network node forwards the segment sequence to the node according to the shortest path calculated by the SPF (if there is an equivalent path, the ECMP can be implemented), and if the path is a Adjacency SID, forwards the segment sequence to the next node according to the Adjacency SID, the data packet reaches the destination node.
Take the LSP from PE1 to PE2 as an example.
1. The loopback address of the LoopBack1 interface on the Egress PE2 is x.x.x.x/x, the SID allocated to the address is 10, and the information is flooded to the entire IS-IS domain.
2. All nodes receive the Node SID from PE2 and generate a label forwarding table.
- Incoming label: Local SRGB start value + Released offset value
- Outgoing label: Next hop SRGB start value + advertised offset value
- Next hop of the outbound interface: Next hop of the outbound interface of the shortest path calculated by IGP
3. The Ingress PE1 performs IS-IS SPF calculation to obtain the two shortest paths from the ECMP to PE2.
The label operation type of the SR is the same as that of the MPLS, including label stack push, label stack swap, and label popping (Pop).
The following figure uses SR-BE Tunnel on PE1 to PE2 as an example.
1. PE1 on the ingress node: After receiving a service packet, the encapsulates the
outgoing label allocated by the next hop P according to the destination address, label forwarding table, and ECMP traffic distribution algorithm, and then forwards the packet through the corresponding interface based on the destination prefix SID and label forwarding table.
Example:
Push 210 to P1
Push 310 to P2
2. Transit node: According to the outer label and the label forwarding table, the outgoing label of the SR is allocated to the next hop, and the outgoing label is forwarded through the corresponding interface.
Example:
Swap 210 to 410 on P1
Swap 310 to 510 on P2
3. Egress node PE2: If the outer label is itself, the outer label is popped and the service packet is forwarded to the CE according to the outer IP address.
Example:
Pop 610 from P3
Pop 610 from P4
1. Generation of labels, label forwarding tables, and topology information
Each node on the network uses the IS-IS SR protocol to allocate adjacent SIDs to its adjacencies, floods the IDs to the entire network, and generates label forwarding entries for the local adjacent SIDs allocated to the nodes. In addition, the topology information with the adjacent SID information is generated. In addition, the manually configured SRGB/prefix SID/node SID, SR capability, and SR algorithm are flooded in the IGP domain through IGP packets. Each node runs ISIS SPF to calculate the shortest forwarding path of each node label, and generates a label forwarding table.
2. Report the label and topology information.
Each node in the network reports the topology information with the adjacent SID information to the controller through the BGP Link-State (BGP-LS).
3. Path calculation
The controller uses the Path Computation Element (PCEP) to calculate the label forwarding path.
4. Deliver trails.
The controller delivers tunnel information through PCEP and delivers tunnel attributes to the first node of the tunnel through NETCONF. The first node of the tunnel reports the tunnel status to the controller through PCEP.
5. Create a tunnel.
The first node of the tunnel establishes a SR-TE Tunnel based on the label stack delivered by the controller.
1. Generation of labels, label forwarding tables, and topology information
Each node on the network uses the IS-IS SR protocol to allocate adjacent SIDs to its adjacencies, floods the IDs to the entire network, and generates label forwarding entries for the local adjacent SIDs allocated to the nodes. In addition, the opology information with the adjacent SID information is generated. In addition, the manually configured SRGB/prefix SID/node SID, SR capability, and SR algorithm are flooded in the IGP domain through IGP packets. Each node runs ISIS SPF to calculate the shortest forwarding path of each node label, and generates a label forwarding table.
2. Report the label and topology information.
Each node in the network reports the topology information with the adjacent SID information to the controller through the BGP Link-State (BGP-LS).
3. Path calculation
The controller uses the Path Computation Element (PCEP) to calculate the label forwarding path.
4. Deliver trails.
The controller delivers tunnel information through PCEP and delivers tunnel attributes to the first node of the tunnel through NETCONF. The first node of the tunnel reports the tunnel status to the controller through PCEP.
5. Create a tunnel.
The first node of the tunnel establishes a SR-TE Tunnel based on the label stack delivered by the controller
SR node performs a label operation on the packet according to the label stack corresponding to the SR-TE tunnel in the packet header. Searches for the outbound interface hop by hop based on the top label of the stack to instruct data packets to be forwarded to the destination address of the tunnel. Take the strict path SR-TE Tunnel from PE1 to PE2 as an example.
1. PE1 on the ingress node: After receiving the service packet, the encapsulates the label stack corresponding to the SR-TE Tunnel according to the SR-TE Tunnel determined by the routing policy or tunnel policy configured in the service. If it is determined that the corresponding stack top label 501 is an adjacent SID, the stack top label 501 is popped up, and the remaining part [103, 304, 406] of the label stack is encapsulated into the service packet, and is forwarded out from the corresponding interface PE1->P1 according to the top label and the label forwarding table.
2. Transit node: If the stack top label is the adjacent SID, the stack top label is displayed, and the stack top label and the label forwarding table are forwarded through the corresponding interface.
Take P1 as an example: If it is determined that the stack top label 103 is an adjacent SID, the stack top label 103 is popped up, and the stack top label 103 is forwarded according to the top label 103 and the label forwarding table from the corresponding interface P1->P3.
3. Egress node PE2: The forwards the service packets to the CE according to the outer IP address of the service packets.
The SR performs a label operation on the packet according to the label stack orresponding to the SR-TE tunnel in the packet header. Searches for the outbound interface hop by hop based on the top label of the stack to instruct data packets to be forwarded to the destination address of the tunnel. Take the loose path SR-TE Tunnel from PE1 to PE2 as an example.
1. PE1 on the ingress node: After receiving the service packet, the encapsulates the label stack [1004, 403, 306] corresponding to the SR-TE Tunnel according to the SR-TE Tunnel determined by the routing policy or tunnel policy configured in the service. Encapsulating the SR outgoing label 1004 allocated by the next hop P node according to the outer label and the label forwarding table, and forwarding the SR outgoing label 1004 according to the label forwarding table.
2、 Transit node P2: According to the outer label 1004 and the label forwarding table, the outer label is switched to the outgoing label 1004 allocated to the next hop, and is forwarded through the corresponding interface P2->P4.
3、 Transit node P4: The outer label 1004 is found to be itself, and the outer label 1004 is popped out. Determining that the next layer label 403 is the adjacent SID, popping the label 403, and forwarding the label forwarding table from the corresponding interface P4->P3 according to the label forwarding table.
4. Transit node P3: Judging that the outer label 306 is the adjacency label, pops out the outer label 306, and forwards the outer label 306 from the corresponding interface P3->PE2 according to the label forwarding table.
5. Egress node PE2: The forwards the service packets to the CE according to the outer IP address of the service packets.