OSPF学习笔记

OSPF 学习笔记第一篇

 

1.cost of interface :
http://www.watchguard.com/help/docs/fireware/10/en-US/index_Left.html#CSHID=en-US%2Fdynamicrouting%2Fospf_interface_cost_f.html|StartTopic=Content%2Fen-US%2Fdynamicrouting%2Fospf_interface_cost_f.html|SkinName=Fireware (en-US)

 

2.ospf process ID

OSPF, unlike Border Gateway Protocol (BGP) or Enhanced Interior Gateway Routing Protocol (EIGRP), does not check the process number (or autonomous system number) when adjacencies are formed between neighboring routers and routing information is exchanged. The only case in which the OSPF process number is taken into account is when OSPF is used as the routing protocol on a Provider Edge to Customer Edge (PE-CE) link in a Multiprotocol Label Switching (MPLS) VPN. PE routers mark OSPF routes with the domain attribute derived from the OSPF process number to indicate whether the route originated within the same OSPF domain or from outside it. If the OSPF process numbering is inconsistent on PE routers in the MPLS VPN, the domain-id OSPF mode command should be used to mark that the OSPF processes with different numbers belong to the same OSPF domain.

This means that, in many practical cases, you can use different autonomous system numbers for the same OSPF domain in your network. However, it is best to use consistent OSPF-process numbering as much as possible. This consistency simplifies network maintenance and complies with the network designer intention to keep routers in the same OSPF domain.

In complex routing architectures you might want to have two (or more) completely seperate OSPF routing processes, one serving one set of interfaces and one serving a different set of interfaces.  These processes can be distinguished by means of the process id.  They will operate entirely independently of each other (unless you choose to redistribute routes between them), and will hold completely seperate link-state databases.  They will pass "like ship in the night" without any interference from each other.

 

3 load balance
http://www.cisco.com/en/US/tech/tk365/technologies_tech_note09186a0080094820.shtml

4 Adjacence
OSPF Adjacencies Must match:
Area ID
Subnet mask (unless on point-to-point link, or virtual-link)
Timers
Stub Flag
MTU (this can be overridden though)

 

Whether to become adjacent

        Adjacencies are established with some subset of the router's
        neighbors.  Routers connected by point-to-point networks,
        Point-to-MultiPoint networks and virtual links always become
        adjacent.  On broadcast and NBMA networks, all routers become
        adjacent to both the Designated Router and the Backup Designated
        Router.

        The adjacency-forming decision occurs in two places in the
        neighbor state machine.  First, when bidirectional communication
        is initially established with the neighbor, and secondly, when
        the identity of the attached network's (Backup) Designated

 

        Router changes.  If the decision is made to not attempt an
        adjacency, the state of the neighbor communication stops at 2-
        Way.

        An adjacency should be established with a bidirectional neighbor
        when at least one of the following conditions holds:

        o   The underlying network type is point-to-point

        o   The underlying network type is Point-to-MultiPoint

        o   The underlying network type is virtual link

        o   The router itself is the Designated Router

        o   The router itself is the Backup Designated Router

        o   The neighboring router is the Designated Router

        o   The neighboring router is the Backup Designated Router

结论：只有建立邻接关系的路由间才会把对方的信息写道自己的LSA中，并告诉其他路由器。减少的LSA的数量

5 Flooding in different networks
On broadcast networks, the Link State Update packets are multicast.  The destination IP address specified for the
Link State Update Packet depends on the state of the interface.  If the interface state is DR or Backup, the
address AllSPFRouters should be used.  Otherwise, the address AllDRouters should be used.
On non-broadcast networks, separate Link State Update packets must be sent, as unicasts, to each adjacent neighbor (i.e., those in state Exchange or greater). 
The destination IP addresses for these packets are the neighbors' IP addresses.

 

6 Loopback interface
A loopback interface is generally used by routers or Layer 3 capable switches so that they have a reliable IP interface for routing updates.
In other words, a loopback interface is a special virtual interface created in software.
It has no associated hardware interface and is therefore always up.
One use of such an interface is to simply assign an IP address to the router as a whole rather than to a specific hardware interface.
This allows the assigned IP address to be used to access the router, for example, Telnet, regardless of which real interfaces are up or down.
Loopback interfaces are also used by some protocols.
For example, the Loopback IP address can be used to assign the OSPF Router ID.
There is no limit on the number of loopback interfaces you can create.
Loopback interfaces do not have internal VLAN IDs or MAC addresses.
With Layer 2 Cisco switches like the 3500XL or 2950, a loopback interface is not needed since these switches are not capable of routing between interfaces.
The management VLAN of the switch acts like the loopback interface since they are both logical interfaces.
When you have a Layer 3 switch like a Cisco Catalyst 3550 or 3750, to configure a loopback interface from the config mode, complete these steps:
Switch#configure terminal
Switch(config)#interface loopback 1
Switch(config-if)#ip address
Switch(config-if)#exit
Note:In order to remove the loopback interface, use the no interface loopback 1 in config mode.

 

7 stub network
A stub network is a somewhat casual term describing a computer network, or part of an internetwork, with no knowledge of other networks, that will typically send much or all of its non-local traffic out via a single path, with the network aware only of a default route to non-local destinations. As a practical analogy, think of an island which is connected to the rest of the world through a bridge and no other path is available either through air or sea. Continuing this analogy, the island might have more than one physical bridge to the mainland, but the set of bridges still represents only one logical path.

An enterprise LAN that connects to the corporate network by only one router, or multiple default routers connected to the same logical upstream destination.
A single LAN which never carries packets between multiple routers connected to it. All traffic is to and/or from local hosts. The routers will only route packets into the LAN if it's destined for the LAN, and out from the LAN if it originated on the LAN).
A person, or workgroup, who is connected to an ISP, by only one router, is a stub network with respect to the ISP. This stub network is part of the ISP's autonomous system, discussed below.
An OSPF stub area is one which only has a default route to the rest of the OSPF routing domain. Such an area may have more than one router, but these routers will only know about the default route to the outside.
A stub autonomous system that is connected to only one other autonomous system, through which it gains access to the Internet. This is also called a stub AS, which characterize the great majority of AS connected to the Internet. [1] as of June 30, 2007, there were 224622 routes seen by the APNIC router. These came from 25577 autonomous systems, of which only 74 were transit-only and 22272 were stub/origin-only. 3305 autonomous systems provided some level of transit.

An enterprise LAN that connects to the corporate network by only one router, or multiple default routers connected to the same logical upstream destination.
A single LAN which never carries packets between multiple routers connected to it. All traffic is to and/or from local hosts. The routers will only route packets into the LAN if it's destined for the LAN, and out from the LAN if it originated on the LAN).
A person, or workgroup, who is connected to an ISP, by only one router, is a stub network with respect to the ISP. This stub network is part of the ISP's autonomous system, discussed below.
An OSPF stub area is one which only has a default route to the rest of the OSPF routing domain. Such an area may have more than one router, but these routers will only know about the default route to the outside.
A stub autonomous system that is connected to only one other autonomous system, through which it gains access to the Internet. This is also called a stub AS, which characterize the great majority of AS connected to the Internet. [1] as of June 30, 2007, there were 224622 routes seen by the APNIC router. These came from 25577 autonomous systems, of which only 74 were transit-only and 22272 were stub/origin-only. 3305 autonomous systems provided some level of transit.

 

8 关于Backbone area和Virtual link的理解

Backbone area实际上就是一个连续的ABR组成的连续的网络（也存在单纯的BR），而虚链接是对Backbone area的一个延伸。将一个非骨干域的内部路径当作一个虚拟的骨干域上的网络路径来使用。来自骨干网上的OSPF包以Unicast的方式在虚链接的两端进行发送。而cost是这个域内路径在路由表内的计算值。

非骨干域必须链接到骨干域，通过虚链接，有利于整个系统的快速收敛，便于排错

 

9 OSPF strange time(待补充)

1)RxmtInterval(The time between link state retransmissions, RxmtInterval, is configured for a virtual link.  This should be well over the expected round-trip delay between the two routers hard to estimate for a virtual link; it is better to err on the side of making it too large.)

2)MinLSArrival(the flooding precedure cannot overwrite the newly installed LSA until MinLSArrival senconds have elspsed)

3)InftransDelay(the LSA's LS age must be incremented by InfTransDelay(must be > 0)when it is copied into the outgoing Link State Upadate Packet)

 

10 When an LSA originated?

For all the OSPF router

1) The LS age field of one of the router's self-originated LSAs reachs the value of refresh time(send a maxaged packet)

2) An interface's state changes

3) An network's Designated Router changes(a new Router LSA should be originated, if it becomes the DR, a network LSA should be originated)

4) One of the neighboring routers changes to /from the FULL state