EIGRP routing updates have all the time contained the subsequent hop discipline (much like BGP updates), which was unused till Cisco IOS launch 12.3 when the no ip next-hop-self eigrp AS-number interface configuration command was carried out.
EIGRP doesn’t set the subsequent hop discipline by default. An EIGRP router receiving a routing replace thus assumes that the following hop of the obtained routes is the sending router. This conduct normally works nicely, however prevents site-to-site shortcuts to be established in DMVPN networks, and ends in suboptimal routing in some route redistribution situations.
The next textual content written by Ivan Pepelnjak
was initially printed on
CT3 wiki. That website online turned unreachable in early 2019. We retrieved the unique textual content
from the Web Archive, cleaned it up, up to date it with current info if obligatory,
and republished it on ipSpace.internet weblog
on June 6, 2023
With the no ip next-hop-self eigrp command configured on an interface, an EIGRP router will set the subsequent hop discipline in outgoing routing updates if the IP subsequent hop in its EIGRP topology desk belongs to the IP subnet of the outgoing interface. The subsequent hop discipline is ready in inside in addition to exterior EIGRP routing updates, optimizing site visitors circulate in route redistribution situations.
NBMA Community with Default EIGRP Subsequent-Hop Processing
The default EIGRP subsequent hop processing shall be illustrated in a easy NBMA community with three routers. R1 is the hub router; R2 and R3 are spoke routers. The examples use Body Relay community between them; it may be a DMVPN Section-1 community or a Provider Ethernet service.
EIGRP break up horizon has been disabled on R1 to permit route propagation between R2 and R3. The related components of the R1’s router configuration are included within the following printout:
Disabling EIGRP break up horizon on the hub router
interface Serial1/0
description Hyperlink to FR
ip handle 10.0.0.1 255.255.255.0
encapsulation frame-relay
no ip split-horizon eigrp 1
frame-relay lmi-type ansi
By default, R1 doesn’t set the subsequent hop discipline in outgoing EIGRP routing replace:
After the EIGRP adjacencies are established, the routing desk on R3 accommodates routes to loopback interfaces of R1 and R2. The next-hop discipline within the EIGRP updates despatched by R1 was empty; the following hop of each routes within the R3’s routing desk is thus R1.
Routing desk on R3
R3#present ip route | start Gateway
Gateway of final resort is just not set
10.0.0.0/8 is variably subnetted, 5 subnets, 2 masks
C 10.0.1.3/32 is straight related, Loopback0
S 10.0.0.2/32 [1/0] by way of 10.0.0.1
D 10.0.1.2/32 [90/2809856] by way of 10.0.0.1, 00:05:37, Serial1/0
D 10.0.1.1/32 [90/2297856] by way of 10.0.0.1, 00:05:37, Serial1/0
C 10.0.0.0/24 is straight related, Serial1/0
The EIGRP topology desk printout shows the EIGRP vector metric and subsequent hop particulars. As anticipated, the bandwidth of each routes is similar, however the delay and the hop-count of R2’s loopback interface have been elevated because the routing replace has handed by means of R1.
EIGRP topology desk on R3
R3#present ip eigrp topology 10.0.1.1 255.255.255.255
IP-EIGRP (AS 1): Topology entry for 10.0.1.1/32
State is Passive, Question origin flag is 1, 1 Successor(s), FD is 2297856
Routing Descriptor Blocks:
10.0.0.1 (Serial1/0), from 10.0.0.1, Ship flag is 0x0
Composite metric is (2297856/128256), Route is Inner
Vector metric:
Minimal bandwidth is 1544 Kbit
Complete delay is 25000 microseconds
Reliability is 255/255
Load is 1/255
Minimal MTU is 1500
Hop depend is 1
R3#present ip eigrp topology 10.0.1.2 255.255.255.255
IP-EIGRP (AS 1): Topology entry for 10.0.1.2/32
State is Passive, Question origin flag is 1, 1 Successor(s), FD is 2809856
Routing Descriptor Blocks:
10.0.0.1 (Serial1/0), from 10.0.0.1, Ship flag is 0x0
Composite metric is (2809856/2297856), Route is Inner
Vector metric:
Minimal bandwidth is 1544 Kbit
Complete delay is 45000 microseconds
Reliability is 255/255
Load is 1/255
Minimal MTU is 1500
Hop depend is 2
NBMA Community with Disabled EIGRP next-hop-self
EIGRP subsequent hop calculations are enabled with the no ip next-hop-self eigrp as interface configuration command. After the WAN interface configuration on R1 has been modified (see the next printout)…
Modified EIGRP interface configuration on R1
interface Serial1/0
description Hyperlink to FR
ip handle 10.0.0.1 255.255.255.0
no ip next-hop-self eigrp 1
encapsulation frame-relay
no ip split-horizon eigrp 1
frame-relay lmi-type ansi
… R1 inserts the next-hop IP handle in outgoing EIGRP routing updates if the following hop’s IP handle in its EIGRP topology desk belongs to the IP subnet of the outgoing interface.
Altering the EIGRP subsequent hop processing on an interface clears all EIGRP adjacencies on that interface.
A pattern EIGRP replace packet despatched from R1 to R3 is displayed within the subsequent determine:
After the EIGRP adjacencies between R1 and R2/R3 are reestablished, the next-hop IP handle of R2’s loopback interface in R3’s IP routing desk is the WAN handle of R2 (not R1):
Modified IP routing desk on R3
R3#present ip route | start Gateway
Gateway of final resort is just not set
10.0.0.0/8 is variably subnetted, 5 subnets, 2 masks
C 10.0.1.3/32 is straight related, Loopback0
S 10.0.0.2/32 [1/0] by way of 10.0.0.1
D 10.0.1.2/32 [90/2809856] by way of 10.0.0.2, 00:09:38, Serial1/0
D 10.0.1.1/32 [90/2297856] by way of 10.0.0.1, 00:09:40, Serial1/0
C 10.0.0.0/24 is straight related, Serial1/0
The modified subsequent hop worth can also be displayed within the present ip eigrp topology printout, which signifies that the routing replace has been obtained from R1 (10.0.0.1), however the next-hop discipline within the obtained replace accommodates the IP handle of R2 (10.0.0.2). The delay part of the vector metric and the hop depend discipline have been elevated (as in comparison with the EIGRP route towards R1’s loopback interface) when R1 has propagated the routing replace between R2 and R3.
The rise in EIGRP delay part might trigger suboptimal routing in DMVPN networks with backdoor hyperlinks.
Modified EIGRP topology desk on R3
R3#present ip eigrp topology 10.0.1.2 255.255.255.255
IP-EIGRP (AS 1): Topology entry for 10.0.1.2/32
State is Passive, Question origin flag is 1, 1 Successor(s), FD is 2809856
Routing Descriptor Blocks:
10.0.0.2 (Serial1/0), from 10.0.0.1, Ship flag is 0x0
Composite metric is (2809856/2297856), Route is Inner
Vector metric:
Minimal bandwidth is 1544 Kbit
Complete delay is 45000 microseconds
Reliability is 255/255
Load is 1/255
Minimal MTU is 1500
Hop depend is 2
R3#present ip eigrp topology 10.0.1.1 255.255.255.255
IP-EIGRP (AS 1): Topology entry for 10.0.1.1/32
State is Passive, Question origin flag is 1, 1 Successor(s), FD is 2297856
Routing Descriptor Blocks:
10.0.0.1 (Serial1/0), from 10.0.0.1, Ship flag is 0x0
Composite metric is (2297856/128256), Route is Inner
Vector metric:
Minimal bandwidth is 1544 Kbit
Complete delay is 25000 microseconds
Reliability is 255/255
Load is 1/255
Minimal MTU is 1500
Hop depend is 1
Please notice that it is sensible to make use of third-party subsequent hops solely when the routers can talk with one another however can’t trade routing updates. DMVPN Section 2 networks are a typical instance.
EIGRP Subsequent Hop Processing in Route Redistribution Eventualities
The no ip next-hop-self eigrp interface configuration command impacts inside and exterior EIGRP routes. This impact is straightforward to reveal in a pattern three router community working two EIGRP processes (Determine 4).
The related configuration instructions of the GW router are proven within the subsequent printout:
EIGRP configuration on the GW router
interface FastEthernet0/0
ip handle 10.0.2.2 255.255.255.0
!
router eigrp 1
redistribute eigrp 2
community 0.0.0.0
auto-summary
!
router eigrp 2
redistribute eigrp 1
community 0.0.0.0
auto-summary
AS1 and AS2 belong to 2 completely different EIGRP processes; the no ip split-horizon eigrp interface configuration command is due to this fact not wanted.
IP routing desk on the GW router accommodates EIGRP routes from AS1 and AS2 with right IP subsequent hops.
GW#present ip route | start Gateway
Gateway of final resort is just not set
10.0.0.0/8 is variably subnetted, 4 subnets, 2 masks
C 10.0.2.0/24 is straight related, FastEthernet0/0
D 10.0.1.3/32 [90/156160] by way of 10.0.2.3, 00:01:30, FastEthernet0/0
C 10.0.1.2/32 is straight related, Loopback0
D 10.0.1.1/32 [90/156160] by way of 10.0.2.1, 00:12:40, FastEthernet0/0
The GW router advertises the path to AS1’s loopback handle as an exterior EIGRP path to AS2. With the default EIGRP interface parameters, the next-hop discipline within the EIGRP replace packet is empty and all of the site visitors between AS2 and AS1 flows by means of the GW router:
Routing desk on AS2 with easy route redistribution on GW
AS2#present ip route | start Gateway
Gateway of final resort is just not set
10.0.0.0/8 is variably subnetted, 4 subnets, 2 masks
C 10.0.2.0/24 is straight related, FastEthernet0/0
C 10.0.1.3/32 is straight related, Loopback0
D 10.0.1.2/32 [90/156160] by way of 10.0.2.2, 00:12:14, FastEthernet0/0
D EX 10.0.1.1/32 [170/158720] by way of 10.0.2.2, 00:12:11, FastEthernet0/0
To optimize the site visitors circulate between AS1 and AS2 you must allow the EIGRP subsequent hop calculations with the no ip next-hop-self eigrp command on the GW router. The subsequent hop processing must be enabled for each EIGRP processes:
Enabling third-party subsequent hop on redistribution router
interface FastEthernet0/0
ip handle 10.0.2.2 255.255.255.0
no ip next-hop-self eigrp 1
no ip next-hop-self eigrp 2
After the no ip next-hop-self eigrp command has been configured, the replace packets despatched from GW to AS2 comprise non-zero values within the next-hop discipline:
The subsequent hop for the ten.0.1.1/32 prefix within the IP routing desk of AS2 is modified accordingly; the site visitors from AS2 to AS1 is shipped straight and bypasses the GW router:
Modified IP routing desk on AS2
AS2#present ip route | start Gateway
Gateway of final resort is just not set
10.0.0.0/8 is variably subnetted, 4 subnets, 2 masks
C 10.0.2.0/24 is straight related, FastEthernet0/0
C 10.0.1.3/32 is straight related, Loopback0
D 10.0.1.2/32 [90/156160] by way of 10.0.2.2, 00:12:14, FastEthernet0/0
D EX 10.0.1.1/32 [170/158720] by way of 10.0.2.1, 00:12:11, FastEthernet0/0
The non-zero worth of the next-hop discipline can also be mirrored within the present ip eigrp topology printout:
Modified EIGRP topology desk on AS2
AS2#present ip eigrp topology 10.0.1.1 255.255.255.255
IP-EIGRP (AS 2): Topology entry for 10.0.1.1/32
State is Passive, Question origin flag is 1, 1 Successor(s), FD is 158720
Routing Descriptor Blocks:
10.0.2.1 (FastEthernet0/0), from 10.0.2.2, Ship flag is 0x0
Composite metric is (158720/156160), Route is Exterior
Vector metric:
Minimal bandwidth is 100000 Kbit
Complete delay is 5200 microseconds
Reliability is 255/255
Load is 1/255
Minimal MTU is 1500
Hop depend is 2
Exterior knowledge:
Originating router is 10.0.1.2
AS variety of route is 1
Exterior protocol is EIGRP, exterior metric is 156160
Administrator tag is 0 (0x00000000)