| OSI model by layer |
|---|
| Internet history timeline |
Early research and development:
Merging the networks and creating the Internet:
Commercialization, privatization, broader access leads to the modern Internet: Contents
Examples of Internet services:
|
Intermediate System to Intermediate System (IS-IS, also written ISIS) is a routing protocol designed to move information efficiently within a computer network, a group of physically connected computers or similar devices. It accomplishes this by determining the best route for data through a packet switching network.
The IS-IS protocol is defined in ISO/IEC 10589:2002 [2] [3] as an international standard within the Open Systems Interconnection (OSI) reference design.
In 2005, IS-IS was called "the de facto standard for large service provider network backbones". [4]
IS-IS is an interior gateway protocol, designed for use within an administrative domain or network. This is in contrast to exterior gateway protocols, primarily Border Gateway Protocol (BGP), which is used for routing between autonomous systems. [5]
IS-IS is a link-state routing protocol, operating by reliably flooding link state information throughout a network of routers. Each IS-IS router independently builds a database of the network's topology, aggregating the flooded network information. Like the OSPF protocol, IS-IS uses Dijkstra's algorithm for computing the best path through the network. Packets (datagrams) are then forwarded, based on the computed ideal path, through the network to the destination.
The IS-IS protocol was developed by a team of people working at Digital Equipment Corporation as part of DECnet Phase V.
The Internet Engineering Task Force (IETF) published IS-IS in 1990 [6] , but that RFC was later retracted and marked as historic [7] because it republished a draft rather than a final version of the International Organization for Standardization (ISO) standard, causing confusion.
The protocol was standardized by ISO in 1992 as ISO 10589, for communication between network devices that are termed Intermediate Systems (as opposed to end systems or hosts) by the ISO. The purpose of IS-IS was to make the routing of datagrams possible using the ISO-developed OSI protocol stack called Connectionless-mode Network Service (CLNS). IS-IS was developed at roughly the same time that the Internet Engineering Task Force IETF was developing a similar protocol called OSPF. IS-IS was later extended to support routing of datagrams in the Internet Protocol (IP), the network-layer protocol of the global Internet. This version of the IS-IS routing protocol was then called Integrated IS-IS. [8]
In IS-IS world there is slightly different terminology which comes from ISO wording. Below is the ISO terminology and its counterpart which is widely used in standards and related documentation.
Compared to OSPF, IS-IS has only two circuit types - broadcast (LAN) and P2P. Therefore, designs such as P2MP are unavailable in IS-IS.
IS-IS adjacency can be either broadcast or point-to-point.
From regular TCP/IP world we are used to know that each Layer 3 interface (including loopback) has its own IPv4 or IPv6 address. The most important point is that loopback interface always stays up (unless deleted) compared to physical or logical interfaces.
Therefore, ISO chose a different approach - instead of assigning layer 3 address to each interface, single address is assigned to loopback interface, while other interfaces are considered as unnumbered. This single address is called NET (Network Entity Title).
On a single intermediate system there can be up to 3 NET addresses. This is useful during migration from one area to another.
NET consists of Area, System ID and NSEL. Area itself consists of AFI (Address Family Identifier) and Area ID.
Area can have variable length of 1 - 13 bytes, System ID is 6 bytes and NSEL - 1 byte.
Let's check on an example NET of 49.0100.1921.6821.1138.00. Here,
Let's imagine, that engineer examines L2 or L1 database, or needs to view a specific LSP. Each LSP has LSP ID, consisting of System ID, Pseudonode ID and Fragment ID. Because generally System ID is router's loopback address, remembering which loopback address to which router is not always convenient.
Similar problem is observed in OSPF, when LSDB or specific LSA is checked - they are listed by Advertising router, which is actually an IP. In case of OSPF, in order to overcome difficulty of remembering router IPs or consulting with list, local DNS resolution can be configured. But as it might be understood, this is not very convenient and fast way, especially during troubleshooting ongoing issues.
IS-IS solves this problem in a very elegant manner - in each LSP there is TLV 137, which displays hostname of the originating router. By this means, all routers know hostnames of other routers in the level by examining LSPs. That's why when viewing LSP in L2 or L1 database, they are displayed by hostname, not System ID.
On the other hand, if needed, hostnames and their matching System IDs can be easily seen from IS-IS which keeps their list.
In IS-IS there is concept of areas, but here it works differently from OSPF. First of all, in contrary to OSPF, in IS-IS area is terminated on router, not link.
In IS-IS, backbone area consists of contiguous Level 2 routers. Level 1 areas can be thought as stub areas in OSPF, where very limited reachability information is available. L1/L2 routers act like area border routers (ABRs) between L1 routers and L2 routers by keeping two databases - L1 database and L2 database.
Here is very important role of L1/L2 router - if it is connected to L2 router in another area, then it sets ATT (ATTached bit) in its L1 LSP. L1 routers which receive this LSP (with ATT bit) add default route to originator of this LSP. This is different from OSPF, where ABR generates default route to stub area routers and sends it via LSA 3.
Another difference of router in L1 area in IS-IS from router in stub area in OSPF is that L1 router can inject external routes into area, which travels up to L1/L2 router. With that, it resembles NSSA area in OSPF (where you cannot have external routes from backbone area, but you can inject external routes to NSSA area which are then translated to regular LSA 5 external routes by NSSA ABRs).
However, by default, external L1 routes are not injected from L1 to L2. This can be changed by policy on L1/L2 router, which accepts L1 external routes and originates them into L2.
In case of OSPF, if at some moment it is needed to inject external routes into OSPF domain from stub area, this can be done only by changing area type from stub to NSSA. This will causes tearing down OSPF neighborship. In IS-IS, this happens hit-less, just by adding a new export policy under protocol.
IS-IS LSPs contain specific information, encoded to Attribute block in LSP header, which is 8 bits long. Here are some of the important ones
When IS-IS was initially introduced, TLVs for IS reachability (TLV 2) and IP reachability (TLVs 128 and 130) could have interface metric no more than 63 (6 bits) and total accumulated path metric of no more than 1023 (10 bits).
Obviously, nowadays with higher link speeds and more hops in the path it would be challenging to stay within these limits.
Therefore, 2 new TLVs - TLV 22 for Extended IS reachability and TLV 135 for Extended IP reachability - were introduced. With this, now link metric can be up to 16.7 million (24 bits) and total accumulated path metric can be up to 4 billion (32 bits).
Olderstyle metric is therefore called narrow metrics, while new style metric - wide metrics.
Wide metrics or narrow metrics can be set on level base.
Compared to OSPF, in IS-IS rules and conditions of adjacency formation are much simpler and mainly depend on the router level.
On broadcast networks IS-IS is prone to issue, similar to OSPF, when all routers on the broadcast segment need to form adjacency and exchange LSPs. Therefore, number of LSPs increase in square.
In order to overcome this issue, on each LAN segment a designated intermediate system (DIS) is elected. The router with the highest priority and System ID wins. But, if a new router shows up and has better priority or System ID, then it is elected as a new DIS.
Elected DIS router is a pseudonode, which uses resources (including System ID) of one real router. DIS describes adjacency between routers in the broadcast segment in hub-spoke manner, where DIS is the hub while other routers (including router, promoted to DIS) are spokes.
Pseudonode ID in LSPs, originated from DIS, always have Pseudonode ID field different from zero.
All routers on the LAN segment form adjacency with only DIS and exchanges LSPs with it.
The function of DIS is to send periodic CSNPs on the LAN segment and reply to PSNPs from other routers. In case of DIS failure a new DIS will be elected in the segment. The role of DIS is not as critical as of DR in OSPF. That's why there is no backup DIS (BDIS) elected in IS-IS compared to BDR in OSPF.
IS-IS supports both simple password and MD5 authentication types. In IS-IS, per-level or per-interface authentication is possible.
In addition, to protect from replay attack, IS-IS uses increasing Sequence number in IIH.
Because IS-IS encapsulates its PDUs into Layer 2 frame, it does not depend on Layer 3 protocols, such as IPv4 or IPv6. This is different from OSPF, which uses IPv4. Therefore, when IPv6 came up, adding IPv6 support to OSPF would require re-writing the protocol. That is how OSPFv3 was created.
In case of IS-IS, TLV 232 for IPv6 interface address and TLV 236 for IPv6 reachability were added to support IPv6. And of course, IPv6 needs to be enabled on the interface.
In order to display supported Layer 3 protocols, also called NLPID (Network Layer Protocol ID), TLV 129 is used. Here, IPv4 has code of 0xCC, while IPv6 - 0x8E.
There might be an issue, if IPv4 and IPv6 topologies do not overlap. This could happen due to misconfiguration or intentionally (if some routers between do not support IPv6). For this situations, multi-topology support is added to IS-IS.
TLV 229 was added to display supported multi-topologies, such as IPv4 unicast and IPv6 unicast.
If multi-topology is enabled, IS-IS will calculate separate SPF tree for IPv4 and IPv6. This means twice the resource usage, but from the other side, this prevents traffic blackholing.
When multi-topology is enabled, then IS-IS will use TLV 222 for Multi-topology IS reachability, TLV 235 for Multi-topology IP reachability and TLV 236 for Multi-topology IPv6 reachability.
Depending on the configuration, the router can have either L1, L2 or both L1/L2 databases, against which it runs SPF algorithm.
But there can be situations, when IS-IS router has exactly the same prefix in different level databases, or external and internal. In order to choose best path in this situations, there is a very specific order, in which the route goes from the most preferred to the least preferred:
IS-IS has Hello packets (IIH) which carry information about the router and are used to form adjacency. Another function of hello packets is to detect failure between routers. This can be problematic, if the routers are not directly connected to each other and there is some active equipment between them.
Technically, it is possible to lower hello and hold time intervals to detect failure faster, but this can put unnecessary stress to router.
Instead of this, BFD can be used. Because BFD is running in data plane over UDP, it nearly does not impact main CPU. Also, BFD can provide sub-second failure detection.
IS-IS is also used as the control plane for IEEE 802.1aq Shortest Path Bridging (SPB). SPB allows for shortest-path forwarding in an Ethernet mesh network context utilizing multiple equal cost paths. This permits SPB to support large Layer 2 topologies, with fast convergence, and improved use of the mesh topology. [9] Combined with this is single point provisioning for logical connectivity membership. IS-IS is therefore augmented with a small number of TLVs and sub-TLVs, and supports two Ethernet encapsulating data paths, 802.1ad Provider Bridges and 802.1ah Provider Backbone Bridges. SPB requires no state machine or other substantive changes to IS-IS, and simply requires a new Network Layer Protocol Identifier (NLPID) and set of TLVs. This extension to IS-IS is defined in the IETF proposed standard RFC 6329.