WeOSOpen Shortest Path First Protocol
Introduction
The Open Shortest Path First (OSPF) protocol is a dynamic routing protocol. Utilization of a dynamic routing protocol can help simplifying configuration, since there is no need to for manual configuration of static routes. Further, it also helps improving the robustness of the network, by automatically being able to adapt the routing based on changes in the network topology.
OSPF is a protocol that falls within the group of interior gateway protocols (IGPs). OSPF is an example of a link-state routing protocol. An advantage that link-state routing protocols, like OSPF, have over distance vector routing protocols, such as RIP, is the fast convergence after a topology change and increased scalability.
For example use-cases and example setups, refer to:
Overview
As mentioned OSPF is an example of what is known as a link-state protocol. In a link-state protocol each individual router announce information about their identity (router-id), along with their directly connected networks, and other neighbouring routers. All this information is flooded throughout the OSPF domain, this ensures that each router will gain complete knowledge about every router and link in the entire topology. This is necessary in order to be able to compute the best path (the least cost path) to reach every destination.
Net-A Net-B
.--.-. .--.-.
( ( )__ ( ( )__
(_, \ ) ,_) (_, \ ) ,_)
'-'--`--' '-'--`--'
| |
| |
| |
.----+-----. .----+-----.
| | | |
| Router-A +------------------------+ Router-B |
| | | |
'----+-----' '----+-----'
| |
| |
| |
.----+-----. .----+-----.
| | | |
| Router-C +------------------------+ Router-D |
| | | |
'----+-----' '----+-----'
| |
| |
| |
.--.-. .--.-.
( ( )__ ( ( )__
(_, \ ) ,_) (_, \ ) ,_)
'-'--`--' '-'--`--'
Net-C Net-D
Figure 1: Simple network topology with interconnected routers and networks.
As an example, Router-A in Figure 1 would send out OSPF messages informing other routers about its router-id, its connected networks, i.e., Net-A and the links towards the routers B and C. The other routers would also do the same thing.
OSPF Hierarchy and Areas
Being a link-state protocol OSPF requires routers ti keep a lot of routing information in their database:
-
Routers will keep a database with information of every router and in link in the entire OSPF domain.
-
OSPF routers can also redistribute and keep routing information learnt from external sources (static routes, routes learnt via other routing protocols, etc.).
Therefore, to reduce the burden of keeping state information about the entire OSPF domain on every single router, the domain can be split into OSPF areas. An example is presented in Figure 2, here the routers have been divided into four different areas. When an OSPF network is split into multiple areas, each router will only have full knowledge of the topology within the area it is part of. Routers will also keep summary information about destinations outside their own are, but no additional information on the topologies of other areas.
. . . . . . . . . . . . . . . . . . . . . . . .
. Area 0.0.0.0 .
. .---. (Backbone area) .---. .
. | R | | R | .
. '---' .---. '---' .
. | R | .
. '---' .
. .
.-------. .-------. .-------.
| | . . . . . . .| |. . . . . . . | |
| ABR | | ABR | | ABR |
. . . . . . | | . . . . . .| |. . . . . . | | . . . . . .
. .---. '-------' . .---. '-------' .---. . '-------' .---. .
. | R | . . | R | | R | . . | R | .
. '---' .---. . . '---' .---. '---' . . .---. '---' .
. | R | . . | R | . . | R | .
. '---' . . '---' . . '---' .
. . . . . .
. Area 0.0.0.1 . . Area 0.0.0.2 . . Area 0.0.0.3 .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Figure 2: Sample OSPF hierarchy with a backbone area and three other areas.
Every network that an OSPF router should be aware of must always be associated with an area. If multiple areas are not necessary they will belong to the backbone, area 0.0.0.0.
Areas are organised in a two-level hierarchy. The top of this hierarchy will always be constituted by the backbone, area 0. Since it is limited to two levels all other areas must be connected to the backbone area. Any router that serve networks that are part of different areas are known as an Area Border Router (ABR). Direct connections between areas that do not traverse through the backbone is prohibited.
Virtual Links
Even though all areas must be connected directly to the backbone (area 0), situations could present themselves where it is not physically possible to connect an area directly to the backbone. Because of this OSPF provides a feature referred to as virtual links, that can allow an area to establish a connection to the backbone through a non-backbone area.
Note
Virtual links are currently NOT supported.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. Area 0.0.0.0 . . Area 0.0.0.1 . . Area 0.0.0.2 .
. . . . . .
. .---. . . .---. . . .---. .
. | R | . . | R | . . | R | .
. '---' . . '---' . . '---' .
. .-------. .-------. .
. .---. .---. | | Virtual Link | | .---. .---. .
. | R | | R | | ABR +==================+ ABR | | R | | R | .
. '---' '---' | | | | '---' '---' .
. '-------' '-------' .
. .---. . . .---. . . .---. .
. | R | . . | R | . . | R | .
. '---' . . '---' . . '---' .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Figure 3: An example of a virtual link between area 2 and the backbone, going through area 1.
Area Types
Areas in OSPF can be designated a number of specific area types. The purpose for this is to allow for an even increased amount of control on the routing information that is distributed throughout the OSPF network. The different area types allow to more specifically decide what type of routing information is allowed. Being able to configure the OSPF network to reduce the amount of routing information that is passed to specific areas, can help reduce the size of the link state database and routing table, that is maintained by each individual router. The following area types can be configured:
-
Standard area: This is the default type of all areas added to the OSPF domain, except for area 0. This area type will accept any link update, summary routes, and external routes.
-
Backbone area (area 0): The main area of an OSPF network, all other areas must be connected to this backbone area. All other areas that wants to communicate with each other must do so through this area. Otherwise, it has the same functionality as a standard area.
-
Stub area: This area type do not accept any external network routes outside of the OSPF domain. Hence, the routers inside the area will not store any routing information to external destinations outside of the OSPF network. Instead, the router connected to the backbone (ABR) will advertise a default route,towards itself, into the area so that external network destinations are reachable. Also, no router inside the area is allowed to redistribute routing information learnt from external sources, no autonomous system border router (ASBR) is allowed.
-
Totally stubby area: Behaves in the same manner as a stub area, and in addition this area type does not allow summary routes from other ares inside the OSPF network. Meaning that the routers in the area will not store any routing information to destinations within the OSPF network. Like a regular stub area the ABR will advertise a default route towards itself, so that both destinations external to OSPF and external to the area within the OSPF network, can be reached.
-
Not so stubby area (NSSA): This type of area is very similar to a stub area, with the difference being that it allows ASBR routers inside its area. This makes it possible to advertise external routes through a stubby area.
-
NSSA totally stub area: Is a combination of a Totally stubby area and a NSSA area.
Tip
For some example use-cases on different OSPF areas refer to here.
Router Types
Within an OSPF domain, routers can be divided into a number of different types. Their role and position, within the OSPF domain, will decide what kind of type it will become. The following are the different router classifications that are used within an OSPF network:
-
Internal Router (IR): Are routers where all of their connections belong to the same area.
-
Backbone Router (BR): Are routers that has at least one interface connected to the backbone area. ABR routers are always backbone routers. Routers that only have interfaces connected to the backbone, are also backbone routers.
-
Area Border Router (ABR): Are routers which are attached to multiple different areas. These routers must also be connected to the backbone area (area 0). An ABR summarize topology information of its attached areas and distribute it to the backbone area.
-
Autonomous System Boundary Router (ASBR): Routers that are attached to routers, part of other Autonomous Systems, running a different routing protocol. ASBRs distribute this external routing information into the OSPF domain.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. Area 1 . . Area 0 . . Area 2 .
. . . . . .
. .-------. .-------. .-------. .-------. .
. | | | | | | | | .
. | IR +-------+ ABR +---+---+ ABR +-------+ IR | .
. | | | BR | | | BR | | | .
. '---+---' '-------' | '-------' '---+---' .
. | . . | . . | .
. .---+---. . . .---+---. . . .---+---. . .--.-.
. | | . . | | . . | | . ( ( )__
. | IR | . . | BR | . . | ASBR +-------(_, \ ) ,_)
. | | . . | | . . | | . '-'--`--'
. '-------' . . '-------' . . '-------' . External Network
. . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Figure 4: A small network of a number of routers in different areas, with the purpose to show different OSPF router types.
In Figure 4 above a simple example example is provided to show how the different router types are assigned.
Configuration
The configuration of OSPF is separated into two different parts. The actual router configuration and OSPF related settings for each individual interface.
Router
OSPF is configured under the router section of the global configuration
context:
example:/#> configure example:/config/#> router example:/config/router/#> ospf example:/config/router/ospf/#>
[no] router-id <ID>-
Set the OSFP router identifier, which must be unique within the OSPF domain.
The router ID is a 32-bit value, given in a dotted decimal form
, where a-d are numbers in the range 0-255. Commonly the router ID is set equal to one of the router’s IP addresses. Default: Auto (
no router-id)no-
Use automatically assigned router-id. The router-id will automatically try to be assigned based on the following selection processes:
-
If the lo interface have a statically configured address (other than 127.0.0.1) that address will be used as the router-id.
-
The current highest statically configured ip address for any of the interfaces used by OSPF.
-
If no static address has been configured the first available address, on any of the interfaces used by OSPF, will be selected as the router-id.
-
ID- ID in a dotted decimal form
, where a-d are numbers in the range 0-255.
Note
The router-id will not be sticky if set to auto. If any configuration changes are made the mentioned selection process will be performed again, so it could result in a new router-id to be selected. In order to be sure what the router-id will be set to in any situation, simply specify it manually.
[no] distance <1-255>-
Set the administrative distance to be used for all routes the are learnt via OSPF.
For additional information on routing distance, see this.
Default: 110
no- Reset the distance to the default value.
[no] passive-interface-
Enable/Disable passive-interface, that control if OSPF should be passive on all interfaces by default.
Whether a specific interface should be passive or not, is configurable for each interface.
Default: Active (
no passive-interface)no- Set to active, not passive.
[no] neighbor <IPADDR[,IPADDR,...]>-
Manually define OSPF neighbours.
This may be useful when intermediate switches do not propagate IP multicast, or when using OSPF in NBMA (non-broadcast multiple access) networks.
Note
Remember to set the network type of each interface to
non-broadcastif you want OSPF to send unicast Hello messages on NBMA links.Default: Disabled
no- Remove configured neighbour.
IPADDR- IP address in standard quad-dotted notation, e.g. 192.168.1.1.
[no] network <NETWORK> [area <AREA-ID>]-
Specify a network to announce and an OSPF area the network belongs to.
Default: Disabled, no
networkentries have been configured when first activating OSPF. The backbone area (0.0.0.0) is used as default area.no- Remove any configured network.
NETWORK- IP address in standard quad-dotted notation including netmask, e.g. 192.168.1.1/24.
AREA-ID- The area ID is a 32-bit number, and is entered in dotted decimal form, or as an integer (0..2 32 − 1).
[no] area <AREA-ID>-
Manage area specific settings: stub, route summarization, etc.
Note
Enters a sub-configuration context for area specific settings.
no- Remove custom area settings.
AREA-ID- The area ID is a 32-bit number, and is entered in dotted decimal form, or as an integer (0..2 32 − 1).
[no] redistribute <connected|static|rip> [metric <0-16777214>] [metric-type <1|2>]-
Redistribute external routing information into the OSPF domain.
Default: Disabled
no- Remove redistribution of external routing information.
connected- Redistribute routes to directly attached networks, which have not been defined to belong to any OSPF area in the “network” command.
static- Redistribute static routes set with the
ip routecommand. rip- Redistribute routes set by RIP.
metric- The cost of reaching the external AS (autonomous system).
metric-type 1- External type 1 routes, E1 in the routing table, include the cost of the ASBR itself in the LSA.
metric-type 2- External type 2 routes, E2 in the routing table, do NOT include the cost of the ASBR, this is the default.
[no] distribute-default [always] [metric <0-16777214>] [metric-type <1|2>]-
Introduce a default route into the OSPF domain, I.e., announce that this router can reach network 0.0.0.0/0.
Default: Disabled
no- Disable distribution of a default route.
always- Always advertise the default route, regardless if one exist or not. Otherwise, it will only be advertised if it exist.
metric- The cost of reaching the external AS (autonomous system).
metric-type 1- External type 1 routes, E1 in the routing table, include the cost of the ASBR itself in the LSA.
metric-type 2- External type 2 routes, E2 in the routing table, do NOT include the cost of the ASBR, this is the default.
Area Settings
OSPF area specific settings can be configured from a sub-context under the router OSPF configuration:
example:/config/router/ospf/#> area 1 example:/config/router/ospf/area-0.0.0.1/#>
[no] default-cost <0-16777215>-
Set the cost of the default route injected into a stub area. This setting only applies to the ABRs of a stub or NSSA area.
Default: 1
no- Reset to the default value.
[no] range <NETWORK> [advertise|not-advertise|substitute <NETWORK>] [cost 0-16777215]-
Configure inter-area route summarisation, route filtering or route substitution.
Default: Disabled
no- Remove configuration.
NETWORK- IP address in standard quad-dotted notation including netmask, e.g. 192.168.1.1/24.
advertise- Use to aggregate routes (within the area) that match the specified
NETWORKrange, before distributing the routes outside the area. Thus, all routes within the provided range are summarised as a single route, when advertised outside the area. not-advertise- Use to prohibit routes (within the area) matching the specified
NETWORKrange, to be distributed outside the area. Thus, all routes within the provided range are filtered. substitute- Use to substitute routes (within the area) matching the specified
NETWORKrange, with another specifiedNETWORKbefore distributing the routes outside the area. Thus, all routes within the provided range are substituted with anotherNETWORK.
[no] stub [no-summary]-
Configure an area as a stub area.
To create a stub area, all routers in the area (ABRs as well as internal routers) must declare the area as stub.
Default: Disabled (i.e., areas are regular OSPF areas by default)
no- Reset to a regular OSPF area.
no-summary- Configure the area as a totally stubby area, this should be done for all ABRs in the area.
[no] nssa [no-summary]-
Configure an area as a not so stubby area.
To create a nssa area, all routers in the area (ABRs as well as internal routers) must declare the area as nssa.
Default: Disabled (i.e., areas are regular OSPF areas by default)
no- Reset to a regular OSPF area.
no-summary- Configure the area as a NSSA totally stub area, this should be done for all ABRs in the area.
Interface Specific Settings
Interface settings that concern OSPF can be found in the ospf sub-context
located under the individual iface configuration context:
example:/#> configure example:/config/#> iface vlan1 example:/config/iface-vlan1/#> ospf example:/config/iface-vlan1/ospf/#>
[no] passive [auto]-
Control passive-interface setting for OSPF on this interface.
Default: Auto (Follows router ospf
passive-interfacesetting)no- Set to active, not passive.
auto- Follow router ospf
passive-interfacesetting.
[no] priority <0-255>-
Configure the designated Router (DR) election priority for this interface. A higher value increases the chance to be elected the designated router.
A priority of 0 indicates that the device is not eligible for election.
Default: 1
no- Reset to the default value.
[no] cost <1-65535>-
Configure the interface OSPF cost.
Default: 10
no- Reset to the default value.
[no] hello-interval <1-65535>-
Configure OSPF hello interval (in seconds) for this interface.
Note
The hello interval setting must be the same on neighbour routers.
Default: 10 (seconds)
no- Reset to the default value.
[no] dead-interval <1-65535>-
Configure OSPF dead interval (in seconds) for this interface.
Note
The dead interval setting must be the same on neighbour routers.
Default: 40 (seconds)
no- Reset to the default value.
[no] network [broadcast|non-broadcast|point-to-point]-
Explicitly set the OSPF network type.
This setting is useful if the OSPF network type for the specific interface needs to be explicitly set. The network type can be set to any of the following explicit modes:
- auto
- broadcast
- non-broadcast
- point-to-point
By default no explicit network type will be selected and the selection performed automatically. This selection is performed separately for each individual interface, based on the type of the interface.
Default: auto (
no network)no- Reset to the default value.
broadcast-
Used when multiple routers are connected to the same broadcast domain, e.g an ethernet LAN network. This network type is useful when all the routers on the network need to be able to communicate with one another.
In auto mode, this will generally be the default for essentially all interface types on the device. The exception to this being any point-to-point interface or loopback.
non-broadcast- If selected it must also be used together with the
neighborsetting in the OSPF router context, i.e, on NBMA links this should be set to send unicast Hello. point-to-point-
This is a network type where the interface is a link between exactly two routers. A packet sent from one router will always have just one recipient.
In auto mode, this will be the default for any point-to-point interface type, e.g. GRE tunnel interfaces.
[no] auth <md5 KEYID | plain> [hash] <SECRET>-
Configure authentication of OSPF messages for this interface.
Two different authentication methods can be selected:
-
MD5: Uses a
KEYIDof 0-255 and aSECRETtext string of 8-16 characters (Both theSECRETand theKEYIDmust be the same on neighbour routers). -
Plain: Uses a
SECRETtext string of 4-8 characters (TheSECRETmust be the same on neighbour routers).
Note
Please note, only one MD5 key currently supported.
Default: Disabled
no- Disable authentication of OSPF messages on the interface.
-
Status
The current status of a OSPF configured router can be observed in the exec context in the following manner:
OSPF Routing Process, Router ID: 20.0.0.2 Supports only single TOS (TOS0) routes This implementation conforms to RFC2328 RFC1583Compatibility flag is disabled OpaqueCapability flag is disabled Initial SPF scheduling delay 0 millisec(s) Minimum hold time between consecutive SPFs 50 millisec(s) Maximum hold time between consecutive SPFs 5000 millisec(s) Hold time multiplier is currently 1 SPF algorithm last executed 7m31s ago Last SPF duration 24 usecs SPF timer is inactive LSA minimum interval 5000 msecs LSA minimum arrival 1000 msecs Write Multiplier set to 20 Refresh timer 10 secs This router is an ASBR (injecting external routing information) Number of external LSA 1. Checksum Sum 0x0000eba6 Number of opaque AS LSA 0. Checksum Sum 0x00000000 Number of areas attached to this router: 1 Area ID: 0.0.0.0 (Backbone) Number of interfaces in this area: Total: 2, Active: 2 Number of fully adjacent neighbors in this area: 2 Area has no authentication SPF algorithm executed 7 times Number of LSA 8 Number of router LSA 4. Checksum Sum 0x00016360 Number of network LSA 4. Checksum Sum 0x00024d06 Number of summary LSA 0. Checksum Sum 0x00000000 Number of ASBR summary LSA 0. Checksum Sum 0x00000000 Number of NSSA LSA 0. Checksum Sum 0x00000000 Number of opaque link LSA 0. Checksum Sum 0x00000000 Number of opaque area LSA 0. Checksum Sum 0x00000000
In addition we could also check the OSPF neighbors:
Router-A:/#> show ip ospf neighbor Neighbor ID Pri State Dead Time Address Interface RXmtL RqstL DBsmL 10.0.14.1 1 Full/Backup 36.044s 10.0.1.2 vlan1:10.0.1.1 0 0 0 10.0.12.1 1 Full/DR 37.909s 10.0.2.1 vlan2:10.0.2.2 0 0 0
We can also check the OSPF database in the following way:
Router-A:/#> show ip ospf database
OSPF Router with ID (10.0.11.1)
Router Link States (Area 0.0.0.0)
Link ID ADV Router Age Seq# CkSum Link count
10.0.11.1 10.0.11.1 1434 0x8000000c 0x3463 3
10.0.12.1 10.0.12.1 1675 0x8000000c 0x8709 3
10.0.13.1 10.0.13.1 1406 0x8000000b 0xf296 3
10.0.14.1 10.0.14.1 1425 0x80000011 0xbdc6 3
Net Link States (Area 0.0.0.0)
Link ID ADV Router Age Seq# CkSum
10.0.1.1 10.0.11.1 1444 0x80000004 0x1de8
10.0.2.1 10.0.12.1 1675 0x80000004 0xec19
10.0.3.1 10.0.13.1 1666 0x80000004 0xf011
10.0.4.2 10.0.13.1 1436 0x80000004 0xf10c
Routes
The purpose of OSPF is to ensure that we learn routes to reach all of our
intended destinations. To check the routes that the are known and what routes
are active we can use the show ip route command as such:
example:/#> show ip route S - Static | C - Connected | K - Kernel route | > - Selected route O - OSPF | R - RIP | [Distance/Metric] | * - FIB route O 10.0.1.0/24 [110/10] is directly connected, vlan1, 19:45:35 C>* 10.0.1.0/24 is directly connected, vlan1 O 10.0.2.0/24 [110/10] is directly connected, vlan2, 19:45:35 C>* 10.0.2.0/24 is directly connected, vlan2 O>* 10.0.3.0/24 [110/20] via 10.0.2.1, vlan2, 19:44:55 O>* 10.0.4.0/24 [110/20] via 10.0.1.2, vlan1, 19:41:00 O 10.0.11.0/24 [110/10] is directly connected, vlan3, 19:45:35 C>* 10.0.11.0/24 is directly connected, vlan3 O>* 10.0.12.0/24 [110/20] via 10.0.2.1, vlan2, 19:44:55 O>* 10.0.13.0/24 [110/30] via 10.0.1.2, vlan1, 19:41:00 * via 10.0.2.1, vlan2, 19:41:00 O>* 10.0.14.0/24 [110/20] via 10.0.1.2, vlan1, 19:41:00
Any route denoted with an “O” is a route that has been learnt by OSPF. In this case we can see that some routes that OSPF have learnt are not used. The reason for this is that in this case also have connected routes to the same destination. A connected route has a lower administrative distance and will be selected over the OSPF route.
If we are
interested in isolating the routes that OSPF have knowledge of we can use the
show ip ospf route command:
Router-A:/#> show ip ospf route
============ OSPF network routing table ============
N 10.0.1.0/24 [10] area: 0.0.0.0
directly attached to vlan1
N 10.0.2.0/24 [10] area: 0.0.0.0
directly attached to vlan2
N 10.0.3.0/24 [20] area: 0.0.0.0
via 10.0.2.1, vlan2
N 10.0.4.0/24 [20] area: 0.0.0.0
via 10.0.1.2, vlan1
N 10.0.11.0/24 [10] area: 0.0.0.0
directly attached to vlan3
N 10.0.12.0/24 [20] area: 0.0.0.0
via 10.0.2.1, vlan2
N 10.0.13.0/24 [30] area: 0.0.0.0
via 10.0.1.2, vlan1
via 10.0.2.1, vlan2
N 10.0.14.0/24 [20] area: 0.0.0.0
via 10.0.1.2, vlan1
============ OSPF router routing table =============
============ OSPF external routing table ===========