The Cisco Certified Network Professional (CCNP) certification was developed by Cisco in 1997 and has grown in popularity almost as much as the CCNA certification. The first version of the CCNP included the ACRC, CLSC, CMTD and CIT exams. The credential was updated once again in 1999 and given the familiar structure we know it by today: Routing, Switching, Remote Access and Support. Cisco revised the tests again in the early summer of 2002 and gave them a new series of exam numbers: 640-603, 604, 605 and 606, respectively. Then in the late summer of 2002, Cisco dumped the brand new 640-603 exam in favor of the newest exam, 640-901, BSCI. The BSCI exam counts as a core exam towards the CCNP, CCDP and the CCIP certifications.

Having reached my limit of knowledge as a CCNA, I started my trek to CCNP in late 2001 by taking Routing first, followed by Switching, Remote Access and Support. Cisco recommends that you take the exams in this order, and I can assure you that once you've made it past the Routing exam, the other exams are easier.

The main objectives of the Routing exam are implementation and configuration of routing protocols. You'll find a great reference cookbook (including the new IS-IS material) from Cisco here: http://www.cisco.com/univercd/cc/
td/doc/product/software/ios11/cbook/ciproute.htm.

Beyond this information, there are self-study guides and even the official Cisco BSCI course if you prefer instructor-led training. For my original exam study I used the CCNP Preparation Library and CCNP Certification Library from Cisco Press. There's a lot of overlap between the two, but I enjoy studying and wanted to experience both libraries. For an update to the Routing study guide, I recommend this information from Cisco.com: http://www.cisco.com/warp/public/
732/Tech/routing/isis.shtml. Only a few study guides have been released to include the addition of IS-IS and map directly to the new exam objectives. Sybex offers CCNP/CCIP: BSCI Study Guide (http://www.sybex.com/SybexBooks.nsf/booklist/4095), but at the time of this writing, I was unable to confirm any releases from Cisco Press.

The CCNP certification requires knowledge of many networking concepts and specifics as they relate to Cisco products and technologies. The Routing exam covers these concepts from an implementation and configuration perspective. In this article, I address some of the high points to study for the new exam by mapping to the official exam objectives, which you’ll find here: http://www.cisco.com/warp/public/10/
wwtraining/certprog/testing/current_exams/640-901.html

Routing Principals
As a CCNA you’re required to have a grasp of RIP for IP, Cisco’s IGRP and the basics of the link state routing protocols EIGRP, OSPF and maybe BGP. For the CCNP Routing exam, you'd better be prepared for in-depth questions for all of the above and more. As with the previous versions of the exam 640-503 and 640-603, my test gave equal time to all routing protocols with a focus on IS-IS and OSPF.

Understanding the routing process is core knowledge for any network engineer. Routers pass both routed and routing protocol traffic to connect network segments of hosts whether they’re running IP, IPX or Appletalk. To perform their jobs correctly, routers need to know about the different networks in the internetwork. They can learn of these through network connections (directly connected), configured static routes (administratively defined) or dynamic routes (learned routes by running a routing protocol).

The routers maintain routing tables with the entries to various networks. Each entry includes the source of the entry as defined above, a network destination, administrative distance, metric value and the interface or next hop address to reach the network (subnet).

The administrative distance can be defined in several ways but is usually predefined by the type of entry. If the entry is via a routing protocol, there are the default values per routing protocol. RIP for IP uses 120 as its default; IGRP 100; Internal EIGRP 90; OSPF 110; and Internal BGP is 200. If the entry is a configured static route, the distance is 1; if the destination network is via a next hop address, it's 0; if it's linked to an interface, the value is also 0 of a connected route. Metric values on the other hand are calculated by various algorithms such as hop count (the number of routers to pass through), which is used by RIP; bandwidth and delay used by IGRP; a composite metric used by EIGRP; a cost value for OSPF and IS-IS; and path vectors or attributes used by BGP.

RIPv1 and IGRP are classful routing protocols, meaning they always assume the network mask is the same throughout the internetwork. Hence, they’re limited; they can’t support summarized routing information VLSM or CIDR. RIPv2, EIGRP, OSPF, IS-IS, and BGP are all classless routing protocols, meaning they don’t assume the network mask and do support Classless Interdomain Routing (CIDR). EIGRP, OSPF and IS-IS are also link-state routing protocols, meaning they send routing updates only when a change in the network has occurred. EIGRP uses DUAL; OSPF and IS-IS use the Dijkstra algorithm for route calculations. More on this later.

Tip: To display the routing table, show ip route works for routers that are configured on an IP network whereas show ipx route or show appletalk route are used when running the IPX protocol or Appletalk. Debugging can also be critical when routing tables are incomplete or incorrect. Pertinent commands include debug ip rip and debug ip igrp transactions.

Extending IP Addresses
As I stated earlier, EIGRP, OSPF, IS-IS, and BGP are classless routing protocols. They support CIDR, which includes Variable Length Subnet Masks (VLSM), hierarchical addressing and route summarization. These methods along with private addresses and NAT are the means to support IP address depletion on the Internet.

Tip: VLSM works by allowing network designers and engineers to use IP addresses with variable masks because each routing update includes the mask. VLSM knowledge is required to pass this exam.

Many people refer to VLSM as the process of subnetting a subnet. Hierarchical addressing ties directly to VLSM, and it works very much like a phone number, whereas each number isn’t maintained by each phone switch. Finally, route summarization is the last component required to minimize the depletion of IP addresses and is a means to have a single IP address represent a collection of IP addresses. The result of these methods and technologies is to minimize the size of routing tables, thereby reducing protocol traffic passed throughout the internetwork by the routers.

Tip: To pass this exam, you need to know when to use route summarization.

IP helper addresses can benefit a broadcast stricken network of any size or when there is a need to consolidate network services DHCP, DNS or TFTP. This works by allowing you to place such servers more strategically that are central to the network for administrative and traffic-flow design. The ip helper-address address command on the router defines eight UDP broadcast ports: TFTP (69), DNS (53), Time (37), NetBIOS (137), BOOTP server and client (67 and 68), and TACACS (port 49). The traffic is forwarded across the router to a specified subnet where the network services exist. To forward only a selected port type, use the ip forward-protocol udp port command. You can use the commands to forward network service broadcast traffic to a subnet or a selected server on a subnet.

Configuring EIGRP
EIGRP supports many of the same technologies OSPF does, such as VLSM, no limitation of network reachability, better use of network bandwidth for routing updates, plus the support for multiple protocols IP, IPX and Appletalk. EIGRP is much more sophisticated than Cisco’s IGRP and has no real limitations except that it is Cisco proprietary and can only be used by Cisco routers. It’s often referred to as a hybrid routing protocol since it uses the metrics of hop count and those seen in pure link-state routing protocols, neighbor and topology tables. Some of the must-know EIGRP commands include: show ip eigrp neighbors and ip eigrp hello-interval and ip eigrp hold-time (the latter two commands to enable the exchange of hello packets). EIGRP also uses IP multicast, address 224.0.0.10, for updates.

The six steps for a route update between routers is 1) Hello, 2) Update, 3) Topology Table update, 4) ACK, 5) Update and ACK; 6) the routers exchange what they have in their topology tables, then each router updates its own routing table. EIGRP makes use of learned routes using the DUAL algorithm and the advanced metric of Bandwidth and Delay (by default), Reliability, Load, and MTU. My favorite anagram for this is Big Dogs Like Red Meat. The DUAL (Diffusing Update Algorithm) uses distance to select the best route to a destination. This is calculated by adding the cost between the next hop and the destination, which is referred to as Feasible Distance. Successors are chosen much the same way; they have the next lowest-cost route. In case an EIGRP router loses its primary route, it will choose a Feasible Successor to the destination network. Many feasible successors can exist.

EIGRP also supports route summarization, which allows for smaller routing tables and updates while still allowing it to scale to larger networks. The command no auto-summary causes EIGRP to behave less like RIP or IGRP and more like a Hybrid routing protocol with the support of link-state routing. You must also configure for summary routes by using the ip summary-address eigrp as-number address mask command on the interface that will provide the summary to the rest of the network. Load balancing across multiple links can be established with EIGRP by using the maximum-paths number command for up to six equal-cost paths and the variance command for unequal-cost paths. Verifying EIGRP operation would include: show ip eigrp neighbors, show ip eigrp topology, show ip eigrp route, show ip eigrp traffic, and a series of debug commands like debug ip eigrp summary.

Configuring OSPF in a Single Area
OSPF is a vendor-neutral routing protocol and is scalable for large internetworks unlike RIP. OSPF also supports VLSM, no limitation of network reachability, better use of network bandwidth for routing updates, faster routing convergence, and a much smarter path selection criteria method. OSPF-configured routers use the Hello protocol to establish and maintain neighbor relationships using the IP multicast address of 224.0.0.5. The Hello protocol packet contains many things such as Router ID, intervals, neighbors, Area ID, router priority and DR and BDR IP addresses. These values are critical when it comes to discovering, choosing and maintaining OSPF routes.

Tip: The Router ID value is a 32-bit binary number that uniquely identifies the router in the OSPF autonomous system. This value is the highest IP address on any active interface and is also used to break a tie between OSPF routers when selecting the designated router (DR) and backup designated router (BDR).

To configure an OSPF router in a single NBMA (Non-Broadcast Multiple Access) area, you must be familiar with OSPF neighbor subinterface configuration. Here’s where the metal meets the road. As I mentioned at the beginning of this article, my Routing exam included two simulation questions and one of them required the knowledge and skills to configure OSPF in a NBMA network. One interesting note: Unlike the new CCNA exam simulators, the Routing exam simulator did support the use of the (?) for simulator-supported commands. That is, you could type the question mark at any time and it would give you a list of supported commands that included the command you were required to use to solve the configuration issue. The exam notifies you that this is a supported feature (?), but reminds you that it only displays the top-level type commands. What it does display is helpful when you’re working against the clock and your mind goes blank. I tried the (?) command many times during my simulator but only received the necessary help for first-level commands such as show or router ospf. So while the software operates like a real router, because it’s a simulator, it only supports a subset of the actual commands you can type in at the command prompt.

For OSPF configuration in a NBMA mode, the commands used are interface serial number.subinterface-number multipoint and router ospf process-id followed by the network address wildcard-mask area area-id command. In a NBMA mode, OSPF operates very much like it does in a broadcast network where the routers exchange update traffic to identify their neighbors and elect a DR and BDR. Configuration of neighbors is required however with the neighbor ip-address command, and neighbors must belong to the same subnet. Once configured, the OSPF operation can be verified with the commands show ip protocols for routing protocol configuration, show ip route ospf for routing table updates, and show ip ospf neighbor and show ip ospf database.

Interconnecting Multiple OSPF Areas
OSPF in a multi-area internetwork creates some challenges such as frequent SPF calculations, larger routing tables and link-state tables. OSPF was designed to operate under such conditions and adds the use of hierarchical routing and intraarea routing to reduce the SPF calculations and routing table sizes. This also adds to the network new types of OSPF routers, Internal, Backbone, ABR (Area Border Router), and ASBR (Autonomous System Boundary Router). You also have to consider the six LSA types, five area types, and internal vs. external route updates types. (For brevity, I have to cut this topic short since an entire article still may not do it justice.)

Tip: There’s so much to learn when it comes to OSPF, that it should probably have an exam of its own. Take the information in stride and study it many times over. Refer to multiple study resources and don’t forget the hands-on.

Configuring IS-IS
IS-IS is also a vendor-neutral routing protocol and shares similarities with OSPF and BGP. IS-IS doesn’t have a backbone area like the OSPF area 0. The IS-IS backbone is a contiguous collection of Level 2-capable routers, each of which can be in a different area. An IS-IS routing domain is similar to a BGP autonomous system. A routing domain is a collection of areas under an administration that implements routing policies within the domain. A two-level hierarchy is used to support large IS-IS routing domains. A large domain may be administratively divided into areas. Each system resides in exactly one area. Routing within an area is referred to as Level 1 routing. Routing between areas is referred to as Level 2 routing. A Level 2 Intermediate System (IS) keeps track of the paths to destination areas. A Level 1 IS keeps track of the routing within its own area. For a packet destined for another area, a Level 1 IS sends the packet to the nearest Level 2 IS in its own area, regardless of what the destination area is. Then the packet travels via Level 2 routing to the destination area, where it may travel via Level 1 routing to the destination.

Tip: Be sure you understand the link cost calculation used by IS-IS, L1 and L2 router types, and the format of NSAP addresses.

To enable IS-IS and specify the area for each instance of the IS-IS routing process on a Cisco router, the commands router isis [area tag], net network-entity-title, interface interface-type interface-number, ip router isis [area tag], and ipaddress ip-address-mask are required to assign the routing process to an interface instead of a network.

Routers running IS-IS will send hello packets out all IS-IS-enabled interfaces to discover neighbors and establish adjacencies if their hello packets contain information that meet the criteria of matching authentication, IS-type and MTU size. Routers may build a link-state packet (LSP) based upon their local interfaces that are configured for IS-IS and prefixes learned from other adjacent routers and all routers will construct their link-state database from these LSPs. Then a shortest-path tree (SPT) is calculated by each IS, and from this SPT the routing table is built.

Tip: IS-IS also supports multi-area design routing. You should be familiar with the operation and basics of configuration.

Configuring Basic BGP
For many, border gateway protocol (BGP) has been one of the greater challenges on the Routing exam. The study resources and features surrounding BGP seem limitless. I had a vague understanding of BGP before my CCNP studies, and I’m still learning. Perhaps after years of working for a service provider or in an enterprise network, one could master the depth of BGP. This area is where I experienced my other router simulator-based question. Fortunately, the simulator only required basic BGP configuration knowledge. Since the simulator supported the (?) command, I felt confident in my answers (more on that later).

BGP is used to connect larger networks that make up the backbone of the Internet by connecting autonomous systems. Each BGP design engineer must apply for his or her own. The AS (Autonomous System) numbers range from 1 to 65536 and the range between 64512 and 65535 are reserved for private use. BGP’s uses are specific. Unless you have good understanding, multiple connections to the Internet or plenty of bandwidth, it’s recommended that you use the ip route prefix mask address/interface distance command to create static routes when your network doesn’t meet the above requirements. BGP uses many of the familiar terminology as mentioned for OSPF such as internal routing and neighbors or peers. BGP between peers can be internal in an AS or between two different autonomous systems; this is referred to as external BGP (EBGP). Policy-based routing in BGP allows for definitions of data flow and the exchange of BGP routes by autonomously controlled BGP systems such as by each service provider of the Internet.

There are two types of BGP attributes used when configuring a network: well-known and optional. Of these, there are the AS-path mandatory, next-hop mandatory, local preference and the optional MED and community. The AS-path attribute is used to identify the source of route updates and gets prepended to the route much like a passport would show your travels. The next-hop attribute defines the neighbor responsible for the received update. The local preference attribute provides a preferred path to exit the AS. The MED or metric attribute is exchanged between autonomous systems and indicates the preferred path into the AS.

For the exam, basic BGP configuration knowledge and experience is a must. As I mentioned, my second simulator-based question required me to configure routers within an AS and define the neighbor relationships, for instance, router bgp 65520, neighbor 10.10.0.10 remote-as 65510 and network 10.10.10.10. Once this was done, I was required to complete the configuration from the other side. The aggregate-address command is used to signal the router to summarize BGP routes reducing the routing table sizes and update traffic. For verifying BGP operation, use the show ip bgp, show ip bgp summary, show ip bgp neighbors and debug ip bgp updates commands.

Tip: The best book on BGP is Internet Routing Architectures by Sam Halabi (Cisco Press).

Optimizing Routing Update Operation
This exam objective covers everything else I haven’t mentioned plus topics Cisco didn’t tell you about. One of the more challenging ones is controlling route updates using route maps. Although they remind me of Access lists, there are differences that can cause confusion. Route maps use match commands to allow route updates between routers if the permit command is used or prevent updates if the deny statement is used. There are many different match commands such as as-path, community, interface, ip address and ip next-hop to name a few. Then there is the series of set commands much like that of the match commands, which allow you to configure values of the route map. An example command for BGP is neighbor ip-address/peer-group-name route-map map-name in/out to control neighbor updates. I recommend Cisco Connection Online (CCO) for more information.

Redistribution deserves an article of its own. Cisco wants to be sure you can configure the routing protocols I’ve mentioned to interoperate with each other. There are many reasons to redistribute route updates from one routing protocol to another: migration from IGRP to EIGRP, integrating a RIP and OSPF network, and enabling non-Cisco and Cisco routers to use a common routing protocol. There are the considerations to be aware of when redistributing, routing feedback or loops, incompatible routing information and inconsistent convergence times. Some of the solutions include Seed Metric and modification of the Administrative distance values. You can configure redistribution between RIP for IP and OSPF since they both use the IP protocol stack. Redistribution between IGRP and EIGRP occurs automatically. EIGRP can also work in conjunction with RTMP for Appletalk. Configuration is performed with the command redistribute protocol process-id metric metric-type route-map subnets tag. The passive-interface command can also be used here to prevent updates from exiting an interface but still allow that interface to listen for updates. Finally, the ip default-network command specifies the outside world when different major network numbers are in place.

I’ve heard many CCNPs say that they consider the Routing exam to be the most difficult in the series. No wonder. It covers wide ground and multiple details. If you pass it, you’ve truly achieved something. Good luck!

More Exam Reviews:

• Designing for Cisco Internetwork Solutions (CCDA)
• Access 2002: Beyond the Basics
• Microsoft's 70-214 Exam: Security Workhorse
• Security+: Get with the Basics