ISIS protocol principle and configuration

Summary of premises:

• IS-IS (Intermediate System-Intermediate System): Like OSPF, IS-IS is also an IGP (Interior Gateway Routing Protocol) based on link state and uses SPF algorithm for route calculation. IS-IS was originally a dynamic routing protocol designed by the ISO organization for the OSI seven-layer model, the connectionless network protocol CLNP (similar to IP in the TCP / IP protocol). Because IS-IS has characteristics (convenience and strong extensibility) that OSPF protocol does not, it is migrated from the OSI7 layer model to TCP / IP, which is integrated IS-IS. The IS-IS we mentioned now Most of it refers to integrated IS-IS.(In order to provide routing support for IP, IETF expanded and modified IS-IS in RFC1195 to enable it to be used in both TCP / IP and OSI environments. The revised IS-IS protocol is called integrated IS-IS. Due to the simplicity and scalability of IS-IS, it is currently widely deployed in large ISP networks.)
• OSPF is generally used in the campus network, but it is not absolute. It is also possible to use IS-IS. For the operator network (backbone network), IS-IS is more used because the routing protocol is selected according to the characteristics of the network.
  Campus network features:
  Application-oriented network, mainly for enterprise network users.
  The number of routers is relatively small, the capacity of the LSDB library for dynamic routing is relatively small, and the layer 3 routing domain is relatively small.
  There is the concept of egress routing, which is sensitive to internal and external routing.
  The regional span is not large, the bandwidth is sufficient, and the link state protocol overhead consumes less bandwidth.
  Routing strategies and policy routing applications change frequently and require refined routing operations.
  OSPF's multi-routing type (internal / external), multi-area type (backbone / common / special), excellent cost rules (based on bandwidth settings), and the variety of network types (four types) have gained great advantages in the campus network Play.
  Backbone network features:
  Service-oriented network, established by ISP (Internet Service Provider), and provides network interconnection services for end users.
  Routing scheduling occupies an absolute dominant position, and the number of routers is huge.
  The architecture level is flat and requires IGP as a basic route to serve the upper layer BGP protocol.
  LSDB is large in scale, extremely sensitive to link convergence, and has a high line cost.
  Pursuit of simplicity and efficiency, high scalability, to meet various customer business needs (IPV6 / IPX).
  IS-IS's fast algorithm (PRC has been strengthened), simple message structure (TLV), fast neighbor relationship establishment, high-capacity routing delivery (based on low Layer 2 overhead) and other characteristics have a natural advantage in the backbone network.

Basic principles of IS-IS protocol

• Components of CLNS:(CLNS is not a specific protocol, but similar to a large architecture)
  ①CLNP (similar to the IP protocol of the TCP / IP protocol stack, IP is the TCP / IP transport layer service, CLNP is the transport layer service in the OSI seven-layer model)
  ②IS-IS: routing protocol between intermediate systems, similar to the IP OSPF
  ③ES-IS: Protocol between the host system and the intermediate system, similar to protocols such as ARP and ICMP in IP.
• For integrated IS-IS, it has both TCP / IP protocol stack capabilities and OSI seven-layer model capabilities, so it supports CLNP networks, IP networks, and dual networks of IP networks and CLNP

IS-IS related terms

Acronyms	OSI terminology	IETF terminology
IS	Intermediate System	Router
IT IS	End System	Host
DIS	Designated Intermediate System	DR in OSPF
SysID	System ID	Router ID in OSPF
LSP	Link State PDU	LSA in OSPF
IIH	IS-IS Hello PDU	Hello packets in OSPF
PSNP	Partial Sequence Number PDU	LSR or LSAck packets in OSPF
CSNP	Complete Sequence Number PDU	DD packets in OSPF

Acronyms	OSI terminology	meaning
NSAP	Network Service Access Point	CLNP address + service port
NET	Network Entity Title	Network entity tag (special NSAP, last byte is 0, Router ID + area)

· NSAP: Network service access point, in fact we need to allocate IP address in TCP / IP, but for our CLNP, we only need to create CLNP address. Because our IS-IS is integrated, the CLNP address is also reserved
. NET: Usually in TCP / IP environment, we need to configure an NSAP when we configure integrated IS-IS on the router, then this NSAP, This NSAP is a special NET address, called a network entity tag. Its speciality is that the last byte is 0.

IS-IS address structure

• NSAP is the address used to locate resources in the OSI protocol, which is equivalent to the address of the OSI network protocol CLNP, similar to the concept of IP addresses
• NET is a special kind of NSAP, its last S bit is 0 (SEL = 00)
• The overall NET address is shown in the figure, one is called IDP and the other is called DSP
• There are AFI and IDI in IDP.
• There are high-order DSP, systemID, and selection bits in DSP
• NSAP size range is 8Bytes-20Bytes
  
• This is a very detailed division of the original IS-IS, but if we now configure integrated IS-IS, we only need to pay attention to the area ID part (Area ID), system-ID part and the selection bit (SEL)

IS-IS address example

• 49.0001 stands for Area ID; aaaa.bbbb.cccc stands for System ID, followed by option bit 00 (fixed in IP network)
• It is recommended to observe backward when observing, because the final selection bit 00 is fixed, and the SystemID represented by aaaa.bbbb.cccc is also fixed. Only the Area ID will change. If you look from the front, you may be wrong.
• AFI has specific usage regulations, such as 47.48, 49. Which one should we use? It is recommended to use 49 for experiment, because 49 has local meaning, like IPv4 private address.
• Involving the factor of specifying the router in IS-IS, so we need to consider the naming method of SystemID. In fact, SystemID is 48 bits or 6 bytes. The function is actually very similar to RouterID in OSPF. It is used to Represents a router in the area, so we can also use RouterID to apply, but because RouterID is 4 bytes, that is, 32 bits, so the length is actually less than 48 bits, so we must find a way to fill 12 Bit, so there are the following operations.
  Demo:
  Router-ID 10.1.1.1
  ① Each group becomes 3 numbers
  010.001.001.001
  ② Regroup the numbers into 3 groups, one group is 4 numbers
  0100.0100.1001 Each group is 2 bytes, total 6 bytes, which is the System-ID of this device
  System-ID and RouterID can not conflict, it is the necessary parameter for us to identify the device in the IS-IS area

Network hierarchical routing domain

• We divide the related equipment into different areas, which is a whole IS-IS topology.
• IS-IS allows the entire routing domain to be divided into multiple areas
• A router currently has up to three Area IDs. The configuration of different area IDs is for smooth area merging, division, and conversion. However, the router ’s system ID must be one.
• Unlike OSPF, in IS-IS,A router must belong to an area, not some interfaces belong to an area, other interfaces belong to another area
• There are backbone areas and non-backbone areas in OSPF. IS-IS is also divided into backbone areas and non-backbone areas. The method of distinguishing between backbone areas and non-backbone areas in IS-IS areas is not by area, but by the type of router. For example, in the above figure, the backbone area is composed of L2 equipment and L1 / 2 equipment as the backbone area.

Types of IS-IS routers

• We can define the types of routers, and the area levels of different routers forming areas are different.

• Level-1 Router: Located inside the non-backbone area.
  Features:
  -L1 routers only form neighbor relationships with L1-capable routers (including L1 / L2 routers) in this area.
  -The L1 router only has Level1 LSDB (L1 LSDB) in the area, and the LSDB contains routing information of all L1 routers in the area-the L1
  router passesThe nearest L1 / L2 router ATT BIT field = 1 Generate the default route pointing to this L1 / L2 Luo as the egress route (L1 area is similar to OSPF stub area)Therefore, there are two cases when forwarding. The first one is in this area, then it can be forwarded directly through the route generated by L1 LSDB; the second case is not in this area, then directly through the nearest L1 / L2 router as an export device to connect to the network outside the area(Note that for the nearest L1 / L2 router, sub-optimal routing may occur)

• Level-2 Router: Located in the backbone area
  Features:
  -L2 routers only form neighbor relationships with L2-capable routers (including L1 / L2 routers) in this area.
  -The L2 router has only Level 2 LSDB (L2 LSDB) in the area, and the LSDB contains all routing information between the areas.
  -It accepts packets from other L2 function routers in the area and forwards the packets to L2 in other areas according to the destination address Router (or L2 router forwarded to the same area).
  -Accept messages from L2 routers in other areas and forward them according to the destination address.
  In fact, for the L2 router, it is a bit like the Transit network segment

• L1 / L2 Router: usually located at the boundary of the area
  Features:
  -Can form a neighbor relationship with any level router in the area; and form an L2 neighbor relationship with L2 or L1 / L2 routers adjacent to other areas
  -There may be two levels of chain Road status database; L1 LSDB is used for intra-area routing; L2 LSDB is used for inter
  - area routing -for L1 / L2 routers, it assumes both L1 and L2 responsibilities.
  -For the L1 / L2 router, if it is in the area where the L1 router is located, it is necessary to notify the exit point of the L1 router in this area, that is, when sending the L1 LSP, theATT bitSet to 1 to send to L1 neighbor.

• For our IS-IS, the backbone area actually contains L1 / L2 Router and L2 Router. The area cannot determine the backbone of IS-IS, but is determined by the router type.

• By default, the initial configuration of Huawei routers is L1 / L2 routers

Neighbor HELLO message

• HELLO PDU (Hello protocol data unit): The purpose of
  HELLO packets is to discover neighbors, negotiate parameters and establish neighbor relationships, and later act as keepalive packets.
  Similar to OSPF, IS-IS establishes a neighbor relationship through the interaction of hello messages. But it will be divided into three types of hello messages according to the scene.
  Level-1 IS-IS in the broadcast network uses Level-1 LAN IIH (Level-1 LAN IS-IS Hello), and the destination multicast MAC is: 0180-c200-0014.
  Level-2 IS-IS in the broadcast network uses Level-2 LAN IIH (Level-2 LAN IS-IS Hello), and the destination multicast MAC is: 0180-c200-0015.
  Non-broadcast networks use P2P IIH (point to point IS-IS Hello). But it does not indicate the relevant fields of DIS (Virtual Node).
  IIH packets need to be used to negotiate the size of the packet sent by both neighbors through the padding field.
• Network types supported by IS-IS: (Currently IS-IS only supports these two network types)
  Point-to-point network type (P2P).
  Broadcast multiple access network type (Broadcast Multiple Access). (DIS exists in the broadcast link, that is, to designate an intermediate system. In fact, the function is to create and update pseudo nodes, and send CSNP periodically every 10s in the LAN to flood the LSP.)
  In special environments such as frame relay, you can create The sub-interface supports P2P network types.
• OSPF supports four types of networks, namely point-to-point, point-to-multipoint, broadcast network, and NBMA network

Neighbor relationship establishment

• On the P2P link, it is divided into a two-way handshake mechanism and a three-way handshake mechanism.
  In the two handshake, as long as the router receives the Hello message from the peer, it unilaterally declares the neighbor as up and establishes the neighbor relationship, but there is a risk of single pass.
  The IS-IS Hello PDU that sends P2P three times finally establishes the neighbor relationship, which is the same as the establishment of the neighbor relationship of the broadcast link.
• On the broadcast link, LAN IIH packets are used to perform a three-way handshake to establish a neighbor relationship.
  When the Hello PDU message sent by the neighbor does not have its own system ID, the state machine enters initialized.
  Only the Hello PDUs received from neighbors will be up with their own system ID, eliminating the risk of single-link transmission.
  A neighbor in the broadcast network will elect DIS (virtual node) after it is up. The function of DIS is similar to OSPF DR (designated router).

DIS and the analogy between DIS and DR

• DIS and pseudo node: DIS is a designated intermediate system (Designated IS). Pseudo-node refers to a virtual router created by DIS in a broadcast network.

• The characteristics of DIS:
  · In the broadcast network, DIS needs to be elected, so after the neighbor relationship is established, the router will wait for the interval between two Hello packets before electing the DIS. The Hello message contains the Priority field, and the one with the highest Priority value will be elected as the DIS of the broadcast network. If the priority is the same, the interface with the larger MAC address is elected as DIS. In IS-IS, the DIS interval for sending Hello by default is 1/3 of that of a normal router, while the interval for other non-DIS routers to send Hello is 10 seconds, so the Hello delay of DIS is 10/3.

• The analogy between DIS and DR:
  · Comparison of preference during elections. DIS can also participate in elections with a priority of 0. The default priority is 64, and the higher the priority, the higher the priority. This is similar to the election of DR in OSPF. OSPF has a priority of 0 and does not participate in the election of DR, and the default priority is 1.
  The election process requires a certain amount of time. The OSPF election DR / BDR process is more complicated, and the ISIS election DIS can wait for the interval between two Hello messages. Simple and fast. The election result ISIS has only one DIS, but OSPF has a DR and a BDR for backup. For ISIS, since there is no BDR-type device, it needs to converge the election faster.
  · There is a preemption relationship in DIS. After the election, a new Router will be added to the link in the later period. If the priority is higher than DIS, it is preemptible, but DR is not preemptible.
  After the election is completed, all routers in the ISIS network link are established adjacency relations. In OSPF, DRothers only form a full adjacency relationship with DR / BDR, and there is only a 2-way relationship between DRothers.

• Regarding the role of DIS and DR: When
  performing SPF calculations, they are treated as virtual nodes, simplifying the logical topology of the MA network (same point).
  All are to reduce LSP / LSA flooding (same point).
  In ISIS, the CSNP can also be sent by DIS to synchronize the LSDB (ISIS extended role) of the link.

• The meaning of the pseudo node is actually very simple, because in the LSP we only describe our routing information. In the MA network, we do not describe the topology information. In fact, there will be certain problems, because we may only have interface information in our LSP. But we have no topology information, just like we only have one type of LSA in OSPF is actually not enough in our OSPF, then the role of the pseudo node is actually DIS to notify all the neighbors on the LAN in the LSP of the pseudo node, and OSPF type 2 LSAs are very similar and generate pseudo-node LSPs

• The presence or absence of pseudo nodes will actually affect the cost value in our LSDB. The impact is similar to the following figure
  ↓ No pseudo node
  
  ↓ A carrier with link status information of the pseudo node
  
  
  

• ISIS TLV:
  The meaning of TLV is: type (TYPE), length (LENGTH), value (VALUE). It is actually a data structure, which contains these three fields.
  The advantage of using the TLV structure to construct messages is flexibility and scalability. The use of TLV fixes the overall structure of the message. To add new features, you only need to add a new TLV. There is no need to change the overall structure of the entire message.
  The network topology and routing information are expressed in TLV structure, which makes the flexibility and expansibility of the message greatly exerted.

• LSP PDU (Link State Protocol PDU):
  LSP is similar to OSPF's LSA and carries link state information, including topology and network number.
  Level-1 LSPs are transmitted by Level-1 routers.
  Level-2 LSPs are transmitted by Level-2 routers.
  Level-1-2 routers can transmit the above two LSPs.
  The LSP packet contains two important fields, the ATT field and the IS-Type field. The ATT field is used to identify the route sent by the L1 / L2 router, and the IS-Type is used to indicate whether the IS-IS type that generated the LSP is Level-1 or Level-2 IS-IS.
  The LSP refresh interval is 15 minutes; the aging time is 20 minutes. However, in addition to waiting for 20 minutes, the aging of an LSP also has to wait for a zero aging delay of 60 seconds; the retransmission time of the LSP is 5 seconds.

• SNP PDU (Sequence Number PDU):
  CSNP (Complete Sequence Number PDU) includes summary information of all LSPs in the LSDB, so that synchronization of LSDB can be maintained between adjacent routers.

• PSNP (Partial Sequence Number PDU) contains partial LSP summary information in LSDB, and can request and confirm LSP.

• The CSNP is similar to the OSPF DD message, which transmits a summary of all link information in the LSDB. PSNP is similar to OSPF LSR or LSAck message used to request and confirm part of the link information.

Link state information interaction

• P2P network LSDB synchronization process: After
  establishing a neighbor relationship, RTA and RTB will first send CSNP to the peer device. If the peer LSDB and CSNP are not synchronized, send a PSNP request to obtain the corresponding LSP.
  Suppose that RTB asks RTA for the corresponding LSP, and then sends PSNP to RTA. RTA starts the LSP retransmission timer while sending the LSP requested by RTB, and waits for RTB to send PSNP as an acknowledgment for receiving the LSP.
  If, after the interface LSP retransmission timer expires, RTA has not received the PSNP message sent by RTB as a response, the LSP is retransmitted until the PSB message of RTB is received as an acknowledgement.
• The newly added router in the MA network interacts with the LSDB of DIS synchronously: it is
  assumed that the newly added router RTC has established a neighbor relationship with RTB (DIS) and RTA.
  After establishing the neighbor relationship, RTC sends its LSP to the multicast address (Level-1: 01-80-C2-00-00-14; Level-2: 01-80-C2-00-00-15). In this way, all neighbors on the network will receive the LSP.
  The DIS in the network segment will add the LSP that received the RTC to the LSDB, and wait for the CSNP message timer to expire (DIS sends a CSNP message every 10 seconds) and send a CSNP message to synchronize the LSDB within the network .
  The RTC receives the CSNP message from the DIS, compares its own LSDB database, and then sends the PSNP message to the DIS to request an LSP that it does not have (for example, RTA and RTB LSPs do not).
  After receiving the PSNP message request, the RTB as the DIS sends the corresponding LSP to the RTC to synchronize the LSDB.

IS-IS routing algorithm

• SPF is currently applied to various routing protocols based on link state. It uses Dijkstra algorithm to calculate the shortest path. The basic idea is to build an SPT (Shortest Path Tree, shortest path spanning tree) based on the topology information described in the LSDB, and then use the routing information described in the LSDB as leaves on the tree to generate the final route.
  iSPF (Incremental SPF, enhanced SPF algorithm) is mainly used when some topologies have changed. In this case, it is not necessary to recalculate the entire network topology, but only a small amount of changed topology is corrected, thereby greatly saving routing calculation time .
  PRC (Partial Route Calculate) is mainly used when only the routing information changes. In this case, there is no need to recalculate the network topology, but just generate new routing information based on the original topology. Greatly save routing calculation time.
  In routing protocols, OSPF and ISIS use SPF algorithm. OSPF integrated the ispf feature after version 12.3. Need to open manually. ISIS seems to have PRC characteristics by default (PRC relies on the shortest path tree calculated by ISPF in the huawei documentation).
  These two characteristics are an improvement to the SPF algorithm.
  In general, PRC is used to deal with the situation where the network topology is unchanged and the routing information changes, and iSPF is used to deal with the change in the network topology (the structure of the shortest path tree) Case

Differences between IS-IS and OSPF

Network type and cost method:
IS-IS protocol only supports two network types, and the default cost value of all bandwidths is the same, OSPF protocol supports four network types, and according to different bandwidths Set the corresponding cost value, and have good support for network types such as frame relay and on-demand link.
Area type: The
IS-IS protocol is divided into L1 / L2 areas. The L2 area is the backbone area with all the detailed routes. L1 has only the default route to L2. The OSPF protocol is divided into backbone areas, common areas, and special areas. Common areas and special areas need to pass through the backbone area for cross-area access.
Packet type:
IS-IS protocol routing bears only LSP packets and the internal routing information does not distinguish between internal and external, simple and efficient, without recursive calculation. OSPF protocol routes carry messages with various LSA types, such as 1/2/3/4/5/7. The routing level is strict and requires recursive calculation, which is suitable for fine scheduling calculation.
Routing algorithm: When
the network segment on a node in the ISIS protocol area changes, the PRC algorithm is triggered, which converges faster and the packet overhead for calculating the route is relatively small. The OSPF protocol participates in the topology construction due to the network address. When the network segment address changes in the area, the i-spf algorithm is triggered, which is relatively complicated and complicated.
Extensibility:
ISIS protocol uses TLV to transmit any routing information. The structure is simple and easy to extend. For example, the support for IPv6 can be solved by adding only 2 TLVs. And ISIS itself supports IPX and other protocols. The OSPF protocol itself is specifically developed for IP, and the OSPF protocols that support IPv4 and IPv6 are two independent versions (OSPFv2 and OSPFv3).

Summary: Therefore, whether to use IS-IS or OSPF should actually be selected according to the specific network characteristics. There is no absolute statement.