Network Bulls
www.networkbulls.com
Best Institute for CCNA CCNP CCSP CCIP CCIE Training in India
M-44, Old Dlf, Sector-14 Gurgaon, Haryana, India
Call: +91-9654672192
As mentioned in the introduction to this chapter, keeping routing tables small helps conserve
memory and may improve the time required by a router to forward packets. Route
filtering allows an engineer to reduce the size of the routing table, but with the side effect
of limiting the destinations reachable by each router. That effect may or may not be acceptable,
given the other design goals of a particular internetwork, and given the need to
operate the network.
Route summarization allows an engineer to keep the routing tables more manageable,
without limiting reachability. Instead of advertising routes for every subnet, a router advertises
a single route that represents the same range of IP addresses as more than one subnet.
Each router can forward packets to the same set of destinations, but the routing table is
smaller. For example, instead of advertising routes 10.11.0.0/24, 10.11.1.0/24, 10.11.2.0/24,
and so on–all subnets up through 10.11.255.0/24–a router could advertise a single route
for 10.11.0.0/16, which includes the exact same range of addresses.
This section begins by examining some design issues related to route summarization.
Then the text moves on to explain how to explicitly configure EIGRP summary routes,
finishing with a discussion of automatically created summaries based on the auto-summary
command and feature.
Route Summarization Design
Route summarization works best when the subnet planning process considers route summarization.
To accommodate summarization, the engineer assigning subnets can assign
www.CareerCert.info
Chapter 4: EIGRP Route Summarization and Filtering 115
10.11.0.0/16
(Best Route
Through WAN1)
10.12.0.0/16
(Best Route
Through WAN2)
10.1.0.0/16
WAN Links
WAN1
WAN2
10.17.32.0/19
Manufacturing
10.16.0.0/16
Data Center
Bx
B1
B2
10.9.1.0/24
Core Links
Core1
Core2
Figure 4-4 Address Blocks Planned for Example Enterprise Internetwork
larger address blocks to one part of the topology. The engineers working with that part of
the internetwork can break the address blocks into individual subnets as needed. At the
edge of that part of the network, the engineers can configure route summaries to be advertised
to the other parts of the internetwork. In short, when possible, plan the route
summaries before deploying the new parts of an internetwork, and then assign addresses
to different parts of the internetwork within their assigned address blocks.
For example, consider Figure 4-4, which shows a variation on the same internetwork
shown earlier in this chapter, with the address blocks planned before deployment.
Figure 4-4 shows the address blocks planned for various parts of the internetwork, as
follows:
■ Assign branch subnets come from two consecutive ranges–10.11.0.0/16 and
10.12.0.0/16.
■ Assign WAN router-to-router subnets from the range 10.1.0.0/16.
■ Assign core LAN router-to-router subnets from the range 10.9.0.0/16.
■ Assign Data Center subnets from the range 10.16.0.0/16.
■ Give the manufacturing division, which has a separate IT staff, address block
10.17.32.0/19.
Inside each of the circles in Figure 4-4, the engineering staff can assign subnets as the
need arises. As long as addresses are not taken from one range and used in another part of
the internetwork, the routers at the boundary between the regions (circles) in Figure 4-4
can configure EIGRP route summarization to both create one large summary route and
prevent the advertisement of the smaller individual routes.
www.CareerCert.info
116 CCNP ROUTE 642-902 Official Certification Guide
Calculating Summary Routes
Note that the examples in this chapter generally use simpler examples of summary routes,
using prefix lengths like /24 and /16 most often. However, for the exam, you need to be
comfortable interpreting prefix/prefix length pairs, and subnet/mask pairs, whether they
represent an actual subnet or a summary route.
The math to analyze a subnet/mask pair, or prefix/length pair, is identical to the math included
as part of the CCNA certification. As such, this book does not attempt to explain
those same concepts, other than this brief review of one useful shortcut when working
with potential summary routes.
If you can trust that the subnet/mask or prefix/length is a valid subnet or summary, then
the following method can tell you the range of numbers represented. For example, consider
10.11.0.0/16. Written in subnet/mask form, it is 10.11.0.0/255.255.0.0. Then, invert the
mask by subtracting the mask from 255.255.255.255, yielding 0.0.255.255 in this case.
Add this inverted mask to the subnet number (10.11.0.0 in this case), and you have the high
end of the range (10.11.255.255). So, summary 10.11.0.0/16 represents all numbers from
10.11.0.0–10.11.255.255.
When using less obvious masks, the process works the same. For example, consider
10.10.16.0/20. Converting to mask format, you have 10.10.16.0/255.255.240.0. Inverting the
mask gives you 0.0.15.255. Adding the inverted mask to the subnet number gives you
10.10.31.255, and a range of 10.10.16.0–10.10.31.255.
Before closing this short section about calculating summary routes, note that the the
process of adding the inveretd subnet mask assumes that the prefix/length or subnet/mask
is a valid subnet number or valid summary route. If it is not, then you can still do the
math, but neither the low end nor high end of the range is valid. For example,
10.10.16.0/19, similar to the previous example, is not actually a subnet number. 10.10.16.0
would be an IP address in subnet 10.10.0.0/19, with range of addresses
10.10.0.0–10.10.31.255.
Choosing Where to Summarize Routes
EIGRP supports route summarization at any router, unlike OSPF, which requires that summarization
be performed only at area border routers (ABR) or autonomous system border
routers (ASBR). EIGRP’s flexibility helps when designing the internetwork, but it also
poses some questions as to where to summarize EIGRP routes.
In some cases, the options are relatively obvious. For example, consider the 10.17.32.0/19
address block in manufacturing in Figure 4-4. The manufacturing division’s router could
summarize all its routes as a single 10.17.32.0/19 route when advertising to Core1. Alternately,
Core1 could summarize all those same routes, advertising a summary for
10.17.32.0/19. In either case, packets from the rest of the internetwork shown in Figure 4-4
will flow toward Core1 and then to the Manufacturing division.
Next, consider the 10.16.0.0/16 address block in the Data Center. Because all these subnets
reside to the right of Layer 3 switches Core1 and Core2, these two devices could summarize
10.16.0.0/16. However, these routes could also be summarized on WAN1/WAN2 for
advertisement to the branches on the left. Summarizing on Core1/Core2 helps reduce the
www.CareerCert.info
Chapter 4: EIGRP Route Summarization and Filtering 117
WAN1
B2 WAN2
B1
BW 768 EIGRP
BW 768
BW 256
BW 256
EIGRP
Summaries:
10.11.0.0/16
10.12.0.0/16
3,000,000
10,000,000
Summaries:
10.11.0.0/16
10.12.0.0/16
10,000,000
3,000,000
Destination
10.11.0.0/16
10.12.0.0/16
Next hop
WAN1
WAN2
Routing Table
Destination
10.11.0.0/16
10.12.0.0/16
Next hop
WAN1
WAN2
Routing Table
1 2
1
2
Core1
Core2
Figure 4-5 Choosing Locations for Route Summarization
size of the routing tables on WAN1 and WAN2. However, the sheer number of subnets in
a Data Center is typically small compared to the number of small remote sites, so the savings
of routing table space may be small.
One advantage of summarizing 10.16.0.0/16 onWAN1/WAN2 instead of Core1/Core2 in
this case is to avoid routing inefficiencies in the core of the internetwork. The later section
“Suboptimal Forwarding with Summarization” discussed the concept with a different
example.
Influencing the Choice of Best Route for Summary Routes
Often, engineers plan route summarization for the same address block on multiple routers.
Such a design takes advantage of redundancy and can be used to perform basic load balancing
of traffic across the various paths through the internetwork. Figure 4-5 shows one
such example, with Routers WAN1 and WAN2 summarizing routes for the two address
blocks located on the branch office LANs: 10.11.0.0/16 and 10.12.0.0/16.
The figure shows the advertisements of the summary routes. WAN1 and WAN2 both advertise
the same summarizes: 10.11.0.0/16 for some branches and 10.12.0.0/16 for the others.
Note that by advertising the WAN routes, instead of filtering, the operations staff
might have an easier time monitoring and troubleshooting the internetwork, while still
meeting the design goal of reducing the size of the routing table. (Also, note that Router
WAN1 summarizes Manufacturing’s routes of 10.17.32.0/19.)
In some cases, the network designer has no preference for which of the two or more routers
should be used to reach hosts within the summary route range. For example, for most Data
www.CareerCert.info
118 CCNP ROUTE 642-902 Official Certification Guide
Center designs, as shown earlier in Figure 4-4, the routes from the left of the figure toward
the Data Center, through Core1 and Core2, would typically be considered equal.
However, in some cases, as in the design shown in Figure 4-5, the network designer wants
to improve the metric of one of the summary routes for a single address block to make
that route the preferred route. Using 10.11.0.0/16 as an example, consider this more detailed
description of the design:
■ Use two PVCs to each branch–one faster PVC with 768 Kbps CIR and one slower
PVC (either 128 Kbps or 256 Kbps CIR).
■ Roughly half the branches should have a faster PVC connecting to Router WAN1, and
the other half of the branches should have a faster PVC connecting to Router WAN2.
■ Assign user subnets from the range 10.11.0.0/16 for branches that use WAN1 as the
primary WAN access point, and from 10.12.0.0/16 for the branches that use WAN2 as
primary.
■ Routing should be influenced such that packets flow in both directions over the faster
WAN link, assuming that link is working.
This design requires that both directions of packets flow over the faster PVC to each
branch. Focusing in the outbound (core-toward-branch) direction for now, by following
the design, and setting the interface bandwidth settings to match the PVC speeds, the outbound
routes will send packets over the faster PVCs. The main reason for the route choices
is the following fact about summary routes with IOS:
Set the summary route’s metric components based on the lowest metric route upon
which the summary route is based.
By setting the interface bandwidth settings to match the design, the two WAN routers
should summarize and advertise routes for 10.11.0.0/16 and 10.12.0.0/16, advertising these
routes toward the core–but with different metrics.
WAN1 advertises its 10.11.0.0/16 route with a lower metric than WAN2’s summary for
10.12.0.0/16 because all of WAN1’s routes for subnets that begin 10.11 are reachable over
links set to use 768 Kbps of bandwidth. All WAN1’s links to branches whose subnets begin
10.12 are reachable over links of speed 128 KBps or 256 Kbps, so WAN1’s metric is
higher that WAN2’s metric for the 10.12.0.0/16 summary. WAN2 follows the same logic
but with the lower metric route for 10.12.0.0/16.
As a result of the advertisements on WAN1 and WAN2, the core routers both have routing
table entries that drive traffic meant for the faster-through-WAN1 branches to WAN1, and
traffic for the faster-through-WAN2 branches to WAN2.
Suboptimal Forwarding with Summarization
The final concept to consider when summarizing routes is that the packets may take a
longer path than if summarization is not used. The idea works a little like this story. Say
you were travelling to Europe from the USA. You knew nothing of European geography,
other than that you wanted to go to Paris. So, you look around and find hundreds of
flights to Europe and just pick the cheapest one. When you get to Europe, you worry
Step 2. WAN1, which has routes for all the subnets that begin 10.11, has a route for
10.11.1.0/24 with WAN2 as the next hop (because WAN1’s link to B1 has
failed).
Step 3. WAN2 has a route for 10.11.1.0/24, with B1 as the next hop, so WAN2 forwards
the packet.
Step 4. B1 forwards the packet to host 10.11.1.1.
Route Summarization Benefits and Trade-Offs
The previous section showed details of a classic trade-off with route summarization: the
benefits of the summary route versus the possibility of inefficient routing. For easier
study, the benefits and trade-offs for route summarization are listed here:
Benefits:
■ Smaller routing tables, while all destinations still reachable.
■ Reduces Query scope: EIGRP Query stops at a router that has a summary route that
includes the subnet listed in the Query but not the specific route listed in the Query.
■ EIGRP supports summarization at any location in the internetwork.
■ The summary has the metric of the best of the subnets being summarized.
Trade-offs:
■ Can cause suboptimal routing.
■ Packets destined for inaccessible destinations will flow to the summarizing router before
being discarded.
Configuring EIGRP Route Summarization
The more difficult part of EIGRP route summarization relates to the planning, design, and
analysis of trade-offs as covered in the preceding section. After you have made those design
choices, configuring route summarization requires the addition of a few instances of
the following interface subcommand:
ip summary-address eigrp asn prefix subnet-mask
When configured on an interface, the router changes its logic for the EIGRP Update messages
sent out the interface, as follows:
■ The router brings down, and then back up, all EIGRP neighbors reachable on that interface,
effectively causing neighbors to forget previous topology information, and
listen to new information (when the neighborships recover).
■ When the neighborships recover, the router advertises the summary route, per the ip
summary-address command, assuming the router has at least one route whose address
range is inside the range of the summary route.
■ The router does not advertise the subordinate routes. (The term subordinate route
refers to the routes whose address ranges are inside the range of addresses are defined
by the summary route.)
Routing Descriptor Blocks:
0.0.0.0 (Null0), from 0.0.0.0, Send flag is 0x0
Composite metric is (28416/0), Route is Internal
Vector metric:
Minimum bandwidth is 100000 Kbit
Total delay is 110 microseconds
Reliability is 255/255
Load is 1/255
Minimum MTU is 1500
Hop count is 1
10.1.2.2 (Serial0/0/0.1), from 10.1.2.2, Send flag is 0x0
Composite metric is (11026688/3847936), Route is Internal
Vector metric:
Minimum bandwidth is 256 Kbit
Total delay is 40110 microseconds
Reliability is 255/255
Load is 1/255
Minimum MTU is 1500
Hop count is 3
! Note that the following command lists only routes in the range
! of the summary – 10.16.0.0 – 10.16.255.255.
WAN2#show ip route 10.16.0.0 255.255.0.0 longer-prefixes
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, 23 subnets, 6 masks
D 10.16.2.0/24 [90/156160] via 10.9.1.14, 00:19:06, FastEthernet0/0.12
D 10.16.3.0/24 [90/156160] via 10.9.1.14, 00:19:06, FastEthernet0/0.12
D 10.16.0.0/16 is a summary, 00:14:07, Null0
D 10.16.1.0/24 [90/28416] via 10.9.1.18, 00:19:06, FastEthernet0/1.16
[90/28416] via 10.9.1.14, 00:19:06, FastEthernet0/0.12
D 10.16.4.0/24 [90/156160] via 10.9.1.14, 00:19:06, FastEthernet0/0.12
WAN2#show ip route 10.16.0.0 255.255.0.0
Routing entry for 10.16.0.0/16
Known via “eigrp 1”, distance 5, metric 28416, type internal
Redistributing via eigrp 1
www.CareerCert.info
Chapter 4: EIGRP Route Summarization and Filtering 123
Routing Descriptor Blocks:
* directly connected, via Null0
Route metric is 28416, traffic share count is 1
Total delay is 110 microseconds, minimum bandwidth is 100000 Kbit
Reliability 255/255, minimum MTU 1500 bytes
Loading 1/255, Hops 0
Example 4-5 shows the results only on Router WAN2, but WAN1 will be identically configured
with the ip summary-address command. With only two branch office routers actually
implemented in my lab, WAN2 needs only two ip summary-address commands:
one for the subinterface connected to Router B1, and another for the subinterface connected
to B2. With a full implementation, this same command would be needed on each
subinterface connected to a branch router.
The example also shows how a router like WAN2 uses a summary route to null0. This
route–10.16.0.0/16 with an outgoing interface of null0–causes the router (WAN2) to discard
packets matched by this route. However, as you can see from the end of Example 4-
5, WAN2 also has routes for all the known specific subnets. Pulling all these thoughts
together, when the summarizing router receives a packet within the summary route’s range
■ If the packet matches a more specific route than the summary route, the packet is forwarded
based on that route.
■ When the packet does not match a more specific route, it matches the summary route
and is discarded.
To ensure that the router adds this local summary route, the router uses the administrative
distance (AD) setting of 5. The user may have typed the ip summary-address eigrp 1
10.16.0.0 255.255.0.0 command, without the 5 at the end. Even so, IOS will add this default
AD value as seen in Example 4-5. With an AD of 5, WAN2 will ignore any EIGRPadvertised
summary routes for 10.16.0.0/16–for example, the summary created by
neighbor WAN1—because EIGRP’s default AD for internal routes is 90. In fact, the output
of WAN2’s show ip eigrp topology 10.16.0.0/16 command lists two known routes for
10.16.0.0/16: one to null0 and the other to branch router WAN1 (outgoing interface
S0/0/0.1). WAN1 uses the lower-AD route to null0, which prevents a routing loop. (Note
that this summary route with outgoing interface null0 is often called a discard route.)
Next, consider the results on the branch routers. The following might be reasonable design
requirements that should be verified on the branch routers:
■ Each branch router’s route for 10.16.0.0/16 should use the primary (faster) PVC (see
Figure 4-7).
■ Each branch router should be able to converge quickly to the other 10.16.0.0/16 summary
route without using EIGRP Queries (in other words, there should be an FS route).
Example 4-6 confirms that both requirements are met.
www.CareerCert.info
124 CCNP ROUTE 642-902 Official Certification Guide
Example 4-6 Results of the 10.16.0.0/16 Summary on Routers B1, B2
! Router B1 first !!!!!!!!!!!!!!!!!!!!
B1#show ip route 10.16.0.0 255.255.0.0 longer-prefixes
! lines omitted for brevity
10.0.0.0/8 is variably subnetted, 5 subnets, 3 masks
D 10.16.0.0/16 [90/3847936] via 10.1.1.1, 00:16:53, Serial0/0/0.1
B1#show ip eigrp topology
! lines omitted for brevity
P 10.16.0.0/16, 1 successors, FD is 3847936
via 10.1.1.1 (3847936/28416), Serial0/0/0.1
via 10.1.2.1 (10514688/28416), Serial0/0/0.2
! Router B2 Next !!!!!!!!!!!!!!!!!!!!
B2#show ip route 10.16.0.0 255.255.0.0 longer-prefixes
! lines omitted for brevity
10.0.0.0/8 is variably subnetted, 5 subnets, 3 masks
D 10.16.0.0/16 [90/3847936] via 10.1.2.5, 00:16:44, Serial0/0/0.2
First, on Router B1, the router has an IP route for 10.16.0.0/16, with outgoing interface
S0/0/0.1. Per Figure 4-7, this subinterface indeed connects to the primary PVC. Per the
show ip eigrp topology command, two possible routes for 10.16.0.0/16 are listed; this
command only lists successor and feasible successor routes. Also, note that the FS route’s
RD (28,416) is less than the successor route’s FD (3,847,936), which means the secondary
route indeed meets the feasibility condition.
The reverse is true on Router B2. B2’s best route for 10.16.0.0/16 uses its S0/0/0.2, which
connects to B2’s primary (faster) PVC through WAN2. Although not shown, it also lists its
backup route over the slower PVC as a feasible successor.
The route summarization feature discussed in this section is sometimes referred to as
manual route summarization to contrast it with the term auto summarization. EIGRP auto
summarization is explained next.
Auto-summary
Automatic summarization, also called auto-summary, causes a router to automatically advertise
a summary route under certain conditions, without the use of the ip summary-address
command. When using auto-summary, if a router has interfaces in more than one
Class A, B, or C network, then that router will advertise a single summary route for an entire
Class A, B, or C network into the other classful network, rather than advertise routes
for the individual subnets. The following is a more formal definition:
When a router has multiple working interfaces, and those interfaces use IP addresses
in different classful networks, the router advertises a summary route for each classful
network on interfaces attached to a different classful network.
Key
Topic
www.CareerCert.info
Chapter 4: EIGRP Route Summarization and Filtering 125
10.3.4.0
10.3.5.0
10.3.6.0
10.3.7.0
10.2.1.0
10.2.2.0
10.2.3.0
10.2.4.0
Yosemite
Albuquerque
Seville
172.16.2.0
S0/0
S0/1
172.16.3.0
Mask: 255.255.255.0
172.16.1.0
Which Route to
Network 10.0.0.0
Do I Believe?
Figure 4-8 Discontiguous Network 10.0.0.0
The auto-summary feature first existed as a required feature of classful routing
protocols. By definition, classful routing protocols (RIPv1 and IGRP) do not advertise
subnet mask information. The omission of the subnet mask in routing updates causes several
design problems–in particular, these protocols cannot support variable length subnet
masks (VLSM), route summarization, or discontiguous network designs.
The newer IGPs–EIGRP, OSPF, and RIP-2–are classless routing protocols because they advertise
the subnet mask and support VLSM. However, with auto-summary enabled,
EIGRP acts like classful routing protocols in one specific way: they do not support discontiguous
networks. To support discontiguous networks with EIGRP, simply disable
auto-summary. The rest of this section further defines the terms and the problem, and
shows the solution of disabling auto-summary.
To better understand discontiguous networks, consider this analogy. U.S. residents can appreciate
the concept of a discontiguous network based on the common term contiguous
48, referring to the 48 U.S. states other than Alaska and Hawaii. To drive to Alaska from
the contiguous 48 U.S. states, for example, you must drive through another country
(Canada, for the geographically impaired), so Alaska is not contiguous with the 48 states.
In other words, it is discontiguous.
More formally:
■ Contiguous network: A single classful network in which packets sent between
every pair of subnets will pass only through subnets of that same classful network,
without having to pass through subnets of any other classful network.
■ Discontiguous network: A single classful network in which packets sent between at
least one pair of subnets must pass through subnets of a different classful network.
Figure 4-8 shows a classic example of a discontiguous network 10.0.0.0. Subnets of class A
network 10.0.0.0 exist on the left and the right, with subnets of class B network 172.16.0.0
in the middle of the internetwork. Following the figure, the problem created by the autosummary
feature is described.
The problem is that when EIGRP auto-summarizes routes at the boundary between classful
networks, then routers in other classful networks cannot route packets to all the destiwww.
CareerCert.info
126 CCNP ROUTE 642-902 Official Certification Guide
nations. For example, because both Yosemite and Seville use auto-summary, they both advertise
a route for 10.0.0.0/8 to Albuquerque. Albuquerque may choose one of the two as
the better route–for example, it may choose the route to the left, through Yosemite. However,
in that case, then Albuquerque cannot forward packets to the network 10.0.0.0 hosts
on the right. Even if Albuquerque decided to add both routes to its routing table, the load
sharing typically occurs per destination IP address, not per subnet. So, some packets
might be delivered to the correct host, and others not.
For EIGRP, two solutions exist. First, you could design the network to not use a discontiguous
network. Alternatively, you can just disable auto-summary using the no auto-summary
subcommand inside EIGRP configuration mode. This command affects the behavior
of the router on which it is configured only and tells that router to not advertise a summary
route for the entire classful network. Instead, that router advertises all the subnets, as
if the auto-summary feature did not exist.
Note: The auto-summary and no auto-summary commands have no effect on routers
that connect to a single classful network.
For classful routing protocols, the only solution is to not use discontiguous classful
networks.
Note: Some confusion exists related to EIGRP’s default for auto-summary. Some IOS
documentation claims that EIGRP defaults to use no auto-summary at later IOS releases,
including 12.4T, but experiments show the opposite. You can confirm the actual setting by
looking at the output of the show ip protocols command.
No comments:
Post a Comment