Routing protocol migration
By stretch | Friday, December 18, 2009 at 11:12 a.m. UTC
Each routing protocol on a Cisco router is assigned a default administrative distance (AD). The term is something of a misnomer; rather than affecting any metric calculations within a protocol, administrative distances are simply a way to prefer one routing protocol over another. If a router has learned of a route via both OSPF and EIGRP, for example, it will prefer the route learned via EIGRP regardless of its metric because EIGRP has a lower default AD.
Here is the ubiquitous table of default AD values:
Protocol | AD |
Connected routes | 0 |
Static routes | 1 |
EIGRP Summary | 5 |
External BGP | 20 |
EIGRP | 90 |
OSPF | 110 |
IS-IS | 115 |
RIP | 120 |
Internal BGP | 200 |
Invalid | 255 |
These defaults can be modified to any number between 1 and 255. This can be especially handy when migrating from one routing protocol to another. For example, consider migrating a network (or part of a network) from IS-IS to OSPF. By default, routers will begin to follow OSPF routes as soon as they are learned because OSPF has a lower AD than IS-IS. This means that the entire migration would need to be done within a single maintenance window to avoid disruption, which isn't always practical, particularly on large networks.
In such a scenario, the distance
command can be used under the new routing protocol to set a higher AD than the current routing protocol, ensuring that routes from the current protocol are always given priority.
R1(config)# router ospf 1 R1(config-router)# distance ? <1-255> Administrative distance ospf OSPF distance R1(config-router)# distance 250
Some protocols, like OSPF, allow for more granular manipulation of AD by route type:
R1(config-router)# distance ospf ? external External type 5 and type 7 routes inter-area Inter-area routes intra-area Intra-area routes
There's an important point to consider when applying this strategy to a migration: routes may be summarized differently within the new protocol. Any more-specific routes learned via the new protocol will be favored over less-specific routes known via the current protocol, regardless of their administrative distances. How then can you guarantee that deployment of the new protocol won't alter the current, stable routing topology?
The table above indicates that any routes with an AD of 255 will be considered invalid. We can take advantage of this behavior and set the AD of our new protocol to 255. Here's an example topology of IS-IS network being migrated to OSPF, with OSPF an assigned AD of 250:
R1# show ip route Codes: C - connected, S - static, R - RIP, M - mobile, B - BGP D - EIGRP, EX - EIGRP external, O - OSPF, IA - OSPF inter area N1 - OSPF NSSA external type 1, N2 - OSPF NSSA external type 2 E1 - OSPF external type 1, E2 - OSPF external type 2 i - IS-IS, su - IS-IS summary, L1 - IS-IS level-1, L2 - IS-IS level-2 ia - IS-IS inter area, * - candidate default, U - per-user static route o - ODR, P - periodic downloaded static route Gateway of last resort is not set 10.0.0.0/8 is variably subnetted, 12 subnets, 2 masks C 10.11.0.0/16 is directly connected, Loopback11 O IA 10.14.0.0/16 [250/21] via 10.0.13.3, 00:05:13, FastEthernet0/1 [250/21] via 10.0.12.2, 00:05:13, FastEthernet0/0 O IA 10.12.0.0/16 [250/11] via 10.0.12.2, 00:05:13, FastEthernet0/0 C 10.0.12.0/24 is directly connected, FastEthernet0/0 O IA 10.13.0.0/16 [250/11] via 10.0.13.3, 00:05:13, FastEthernet0/1 C 10.0.13.0/24 is directly connected, FastEthernet0/1 i L1 10.2.0.0/16 [115/20] via 10.0.12.2, FastEthernet0/0 i L2 10.3.0.0/16 [115/20] via 10.0.13.3, FastEthernet0/1 C 10.1.0.0/16 is directly connected, Loopback1 i L2 10.4.0.0/16 [115/30] via 10.0.13.3, FastEthernet0/1 [115/30] via 10.0.12.2, FastEthernet0/0 i L2 10.0.24.0/24 [115/20] via 10.0.12.2, FastEthernet0/0 i L2 10.0.34.0/24 [115/20] via 10.0.13.3, FastEthernet0/1
Note that a number of new routes, not present in IS-IS, have been learned by OSPF (with an AD of 250 versus IS-IS's default AD of 115) and added to the routing table. This can lead to unpredictable and undesired routing. Here's the same table after increasing OSPF's AD to 255:
R1(config)# router ospf 1 R1(config-router)# distance 255 R1(config-router)# ^Z R1# show ip route ... Gateway of last resort is not set 10.0.0.0/8 is variably subnetted, 9 subnets, 2 masks C 10.11.0.0/16 is directly connected, Loopback11 C 10.0.12.0/24 is directly connected, FastEthernet0/0 C 10.0.13.0/24 is directly connected, FastEthernet0/1 i L1 10.2.0.0/16 [115/20] via 10.0.12.2, FastEthernet0/0 i L2 10.3.0.0/16 [115/20] via 10.0.13.3, FastEthernet0/1 C 10.1.0.0/16 is directly connected, Loopback1 i L2 10.4.0.0/16 [115/30] via 10.0.13.3, FastEthernet0/1 [115/30] via 10.0.12.2, FastEthernet0/0 i L2 10.0.24.0/24 [115/20] via 10.0.12.2, FastEthernet0/0 i L2 10.0.34.0/24 [115/20] via 10.0.13.3, FastEthernet0/1
The OSPF-learned routes are gone. This configuration succeeds in protecting our stable routing topology. But if the routes aren't added to the table, how can we verify that OSPF is working correctly at all? By examining the OSPF database, of course:
R1# show ip ospf database OSPF Router with ID (10.1.0.1) (Process ID 1) Router Link States (Area 0) Link ID ADV Router Age Seq# Checksum Link count 10.1.0.1 10.1.0.1 663 0x80000006 0x001E73 2 10.2.0.1 10.2.0.1 657 0x80000005 0x003146 2 10.3.0.1 10.3.0.1 652 0x80000005 0x006CEF 2 10.4.0.1 10.4.0.1 545 0x80000007 0x0079C5 2 Net Link States (Area 0) Link ID ADV Router Age Seq# Checksum 10.0.12.2 10.2.0.1 765 0x80000001 0x00FCFE 10.0.13.3 10.3.0.1 752 0x80000001 0x00E710 10.0.24.4 10.4.0.1 745 0x80000001 0x007078 10.0.34.4 10.4.0.1 746 0x80000001 0x000ECF Summary Net Link States (Area 0) Link ID ADV Router Age Seq# Checksum 10.11.0.0 10.1.0.1 659 0x80000001 0x0024F6 10.12.0.0 10.2.0.1 653 0x80000001 0x001009 10.13.0.0 10.3.0.1 648 0x80000001 0x00FB1B 10.14.0.0 10.4.0.1 541 0x80000001 0x00E72D Router Link States (Area 1) Link ID ADV Router Age Seq# Checksum Link count 10.1.0.1 10.1.0.1 664 0x80000001 0x009B68 1 Summary Net Link States (Area 1) Link ID ADV Router Age Seq# Checksum 10.0.12.0 10.1.0.1 665 0x80000001 0x007E92 10.0.13.0 10.1.0.1 665 0x80000001 0x00739C 10.0.24.0 10.1.0.1 665 0x80000001 0x005E9C 10.0.34.0 10.1.0.1 665 0x80000001 0x00EF01
Posted in Routing
Comments
December 18, 2009 at 4:53 p.m. UTC
Very informative stretch, as usual! I just wanted to add that you can assign different ADs to more specific routes and/or to routes learned from specific sources by using the following syntax:
distance 255 [source IP] [source wildcard] [access list]In this case, the AD will only be applied to routes learned from peers in the "source IP/wildcard" range (in OSPF this would be based on their router ID) and will only apply to routes that match the specific ACL.
December 18, 2009 at 5:11 p.m. UTC
Perhaps add EIGRP external AD 170 as well to the table?
December 18, 2009 at 5:13 p.m. UTC
Hi Stretch,
Perhaps add EIGRP external AD 170 to the table as well?
Cheers, CB
December 25, 2009 at 5:11 a.m. UTC
Not that any real network would ever use it, but ODR has an AD 160....
February 11, 2016 at 6:30 a.m. UTC
Nice article. Very easy to understand.