TCOM 515

Lecture 7

Review for Final• First Course Evaluation

• Any specific questions from the class

• Final outline

• Quick review of the last 3 Lectures hitting the high points

• Quick review of the first 3 Lectures hitting the high points

• I will be in the TCOM Lab on Saturday 5/14 from 12noon to 2pm for office hours and to pick up Lab 6 submissions. I will answer questions via email up until 7pm on Monday 5/16. Final is 5/17 @7:20pm, for those with a final until 7pm, I will allow a late start at 7:50pm, until 10:30pm.

• Website has Final Review document available.

Final OutlineBring a blue book to use for the final.

60 questions worth 1.5 points each including multiple choice, matching and ordering, similar to the midterm = 90 points. 20 questions (1/3) are from material covered before the midterm. 40 questions (2/3) are from material covered since the midterm.

1 question worth 10 points that will involve network design. You will need to a network design in words and tables and turn it into a diagram including routers, links, IP addresses, and routing protocol specific information like area, AS, router type, etc. All routing protocols covered are fair game. The more information the better.

Network Design ProblemYou have the subnet range 10.10.8.0/22. You have 8 routers, each router connects to

2 other routers via Ethernet to form a ring, with 8 connections. Each router requires a loopback. 2 of the routers have additional Ethernet LAN segments off the router that include a switch and 48 hosts, all of which require IP addresses. 2 other routers have Serial connections to two different ISP with an AS of 345 and 456. Our AS is 567. All the routers without LAN segments should have all of their links to other routers without LAN segments in Area 0. All of the routers with LAN links should have their links in a another area, you decide the number. All routers have interfaces of E1/0, E1/1, E1/2, S0.

IS-IS Areas/Levels• Two types of area - defined by levels• Level 2 - like Area 0 of OSPF, form backbone• Level 1 - non-backbone areas• A router can be L1, L2 or L1/L2.• An L1 router only knows about its area.• An L1/L2 router must maintain separate L1 and L2

Link State Databases, L2 routes are not advertised to L1 routers

• An L2 routers, L1/L2 routers and their interconnecting links form the IS-IS backbone.

IS-IS Areas ExampleAreas

L1L2

L1L2

L1

L1

L1

Area 49.001

Area 49.003Area 49.0002

L1L2

Level-1Area

Level-1Area

Level-1Area

Level-2 Backbone

Hierarchical Routing

Area 49.001 Area 49.0002

Level-1Routing Level-2

Routing

Level-1Routing

• IS-IS supports 2-level routing hierarchy

• Routing domain is carved into areas. –Routing in an area is level-1. –Routing between areas is level-2

Backbone

IS-IS Network IDs• Area ID - area address associated with an entire

router, not a link– Each router can have up to three area addresses– Used by Level 2 routing

• System ID - router ID like, uniquely identifies an IS-IS router– Used by Level 1 routing

• NET - Network Entity Title - made up of an Area ID, a System ID and an NSAP Selector– NSAP - Network Service Access Point– RFC 1237 - specifies NET configuration

Update ProcessThe Update process of the routing function is

responsible for constructing the L1 and L2 Link State Databases.

IS-IS Link State Packets (LSPs) - contain values for:– Remaining Lifetime - begins at Max Age and

decrements to zero over time - default 1200 seconds– Sequence Number - starts at 1 and orders the

LSAs from the router– Checksum - Used to check LSA for correct info, if

there is an incorrect checksum, the Remaining Lifetime is set to 0 and flooded throughout the network to purge it from all the routers.

Update Process 2SNP - Sequence Number PDUsPSNP - Partial SNP• Used to acknowledge each LSP as it receives it.• Sent on a Point-to-Point link.• Identifies the LSP by its:

– LSP ID– LSP Sequence Number– LSP Checksum– LSP Remaining Lifetime

• If the LSP is not acknowledged by the timer, the router resends the LSP on the link. The default timer is 5 seconds.

Update Process 3SNP - Sequence Number PDUsCSNP - Complete SNP• Used on broadcast subnetworks.• Designated router multicasts a CSNP.• This CSNP describes every LSP that is currently in

the DR’s Link State Database.• When a router receives a CSNP, it compares the

LSPs described in the CSNP with its own Link State Database.

• If the router has a newer version of an LSP or an LSP not in the CSNP, it multicasts the LSP onto the network.

• If the router has an older version or is missing a copy of an LSP in the CSNP, it updates its database.

IS-IS Route MetricsThe IS-IS route metrics have a value between 0 and 63:Default - supported and understood by every IS-IS

routerDelay - optional metric reflecting transit delay of

subnetworkExpense - optional metric reflecting monetary cost of

using the subnetworkError - optional metric reflecting the residual error

probability of the subnetworkCisco only supports the default metric.Each metric will determine the optimal router for each

destination. So with all four metrics being used that is four SPF calculations for each destination.

IS-IS Decision ProcessThe IS-IS Decision Process uses the Link State database built by the

Update process to calculate the SFP for each destination.A default metric of 10 is assigned to every interface by default but can be

changed. If left at 10, with all interfaces equal, metric becomes a hop count.

Total cost of a route is the sum of metrics of outgoing interfaces along the path. Max is 1023.

L1 and L2 routes are calculated for each destination.L1 path is preferred over an L2 path.Up to six equal costs paths will be put into the routing table, IS-IS will load

balance over the paths.Routes are also classified as internal - destination within the domain or

external - destination not in domain.

BGP Open Message (1)The Open message is the first sent. It is used for identification and

agreement of protocol parameters.Open Message fields:Version - BGP Version number of senderASN - Autonomous System Number - the AS of the sending router,

compared to BGP neighbor configured AS Hold Time - number of seconds that the sender proposes to use as

a hold timer, max time it will wait for keepalive from neighborBGP Identifier - value used to identify the BGP speaker, chosen by

sender, must be unique - Usually IP address of Peer

BGP Update MessageThe Update message is used for most communications between

two BGP peers. Used to advertise or withdraw a route

Update Message fields:Withdrawn Routes - list of IP prefixes for which the sender no

longer wishes to forward packets Path Attributes - list of BGP attributes that describe the prefixes in

the next field - Attribute flag, type code, length and value, for each attribute. Important part of the BGP routing decisions

Network Layer Reachability - list of prefixes advertised and associated with the Path Attributes, all prefixes are described by all path attributes

BGP Notification MessageThe Notification message is used to identify an error in the

underlying TCP connection before it closes the connection.

Notification Message fields:Error Code - identifies the type of error that occurred

1. Message Header Error2. Open Message Error3. Update Message Error4. Hold timer Error5. Finite State Machine Error6. Cease - no other code applies

Error Subcode - narrows down more specific the type of errorData - only present for specific error code and error subcode

combinations

BGP Keepalive• The Keepalive message is used by BGP Neighbors

to maintain that the connection between them is active.

• The hold timer is negotiated at the beginning of the session and used to determine the maximum amount of time between keepalives before a neighbor is considered dead.

• Either a Keepalive message or an Update message will reset the hold timer.

• A Keepalive message is the BG header with no other data contained within it.

BGP RIBs• RIB - Routing Information Base - BGP-4’s term for

the routing table

• There are a few types of RIBs:– Adj-RIB-In - the location where prefixes from specific

neighbors are stored. Each peer has an Adj-RIB-In.– Loc-RIB - all the prefixes in the different Adj-RIB-In are

processed. The chose paths for each individual prefix is stored in the Loc-RIB. Each system has one Loc-RIB.

– Adj-RIB-Out - the location where the prefixes to be advertised to a specific peer are stored. Each peer has its own Adj-RIB-Out.

BGP Origin• The BGP Origin Path Attribute describes how the

advertised prefix came into BGP table at the originating AS.

• The prefix can come from directly attached interfaces, static routes, or other routing protocols.

• The Origin attribute type code is 1 and is a mandatory path attribute.

• The possible values of the attribute are:– 1 - IGP - prefix learned from an IGP– 2 - EGP - prefix learned from an EGP– 3 - Incomplete - learned through method other than IGP or

EGP, most often static route.

BGP Local-Pref• The BGP Local-Pref goes beyond Mutli-Exit-

Discriminator for use by large ISPs with multiple connections and paths to other Autonomous Systems.

• The Local-Pref attribute is configured for the local AS and is assigned based on the advertising AS. The remote AS does not assign the value.

• The higher the Local-Pref value the better the route. It allows the local Network Administrator to choose the preferred paths and change the preference dynamically based on new info, etc.

• This attribute is type code 5 and is well-known and discretionary.

BGP Route Selection• BGP has a very specific process for choosing a

route from overlapping routes.

• Multiple routes to the same prefix mean that the prefix must be of exactly the same subnet length to be considered the same. Routes with more specific subnets or longer subnet lengths are preferred over less specific.

• BGP uses the Attributes we discussed already to help it make decisions when there are equal length prefixes with more than one possible route.

BGP Route Selection• First tie breaker is the highest Local-Pref value.• Second tie breaker is the shortest AS-Path list.• Third tie breaker if configured is the lowest Multi-Exit-

Discriminator value if there are multiple links to the same neighbor.

• Fourth tie breaker is the cost of the path to the Next-Hop subnet. The lower cost wins.

• Fifth tie breaker is that eBGP routes are preferred over iBGP routes.

• Sixth tie beaker is that if all remaining routes are learned via iBGP, the iBGP neighbor with the lowest BGP identifier wins.

Route Redistribution• Route redistribution involves a router using a routing

protocol to advertise routes it learned through another routing protocol, static routes or direction connection.

• Route redistribution is often used on the border routers, to move routes between the IGP and EGP.

• Static routes maybe be needed to advertise into the dynamic routing protocols.

• Route redistribution can be used to bridge legacy networks or connect vendor specific networks.

• Route redistribution is part of an efficient network to connect the LAN to the WAN.

Route Redistribution

• Redistribution of routes between routing protocols that only do classful routing to those that do classless routing and vice versa can behave differently than expected.

• Classful routing protocols may not advertise a given route or may summarize up to the assigned class subnet causing problems and outages.

• Network administrators should only redistribution between classful and classless subnet when necessary and only specific routes needed.

Redistribution Models• There are two network models in which we will discuss the how

and why of route redistribution.– Data Center or Non-transit network– Backbone and/or Transit network

• Redistribution is occurs in one of 3 forms– Static route to a routing protocol– EGP to an IGP– IGP to an EGP

Route Redistribution Types• There are two network models in which we will discuss the how

and why of route redistribution.– Data Center or Non-transit network– Backbone and/or Transit network

• Redistribution is occurs in one of 3 forms– Static route to a routing protocol– EGP to an IGP– IGP to an EGP

Data Center Local Route

• How does a Data Center send its local routes to its ISP?– A static route to NULL0 that is redistributed

into BGP, sent to ISP– The IGP local routes are redistributed into

BGP, sent to ISP– The ISP has static routes for the Data Center’s

local routes configured to point to the interface that connects to the Data Center

ISP Peer Routes

• How can an ISP router tell its Peer about its local routes?– Static routes for all local and Customer routes

that are redistributed into BGP.– IGP routes are redistributed into BGP.– Statics routes configured on Peer for ISP’s local

and Customer routes to Peering interface.– iBGP run between all ISP routers.

ISP Peer Routes

• How can an ISP router tell its other routers about the Peer’s advertised routes?– BGP routes are redistributed into IGP.– Statics routes configured on ISP for Peer’s local

and Customer routes to Peering interface.– iBGP run between all ISP routers.

ISP Customer Routes

• How can an ISP tell its Customers about the rest of the Internet routes?– ISP redistributes Peer and Local routes into

BGP advertisement.– ISP sends default route via BGP advertisement.– ISP has the Customer configure static route for

0.0.0.0 pointing to the interface that connects to the ISP.

ISP Customer Routes

• How can an ISP router tell its other routers about the Customer’s advertised routes?– ISP redistributes Customer BGP routes into its

IGP.– ISP runs iBGP between all routers.– ISP configures static routes for the Customer’s

local routes pointing to the interface that connects to the Customer and redistributes into IGP.

Default Route

• A full Internet routing table contains upward of 160,000 routes. This consumes significant of memory and processing power of a router. Large backbone routers are built to handle routing tables of this size. However smaller routers cannot handle routing tables of this size.

• For networks that have only one connection to the Internet, there is no need to carry a full routing table. Routers only need to know about local subnets and where to send everything that isn’t local.

• Default routes are used as guide to the rest of the world or as a route of last resort. The syntax of a default route is 0.0.0.0

• A default route is the least specific subnet match for a destination and used when there is no other match.

Access ListsAccess list syntaxAccess-list number permit or deny protocol source destination port number extensions

• Number - each access list has a number and each statement in that list has the same number

• Permit or deny - each statement directs the router to either permit or deny the matched packets

• Source - this maybe a single host a /32, a subnet of any size or the word “any” which is a wildcard

• Destination - this maybe a single host, a /32, a subnet of any size or the word “any” which is a wildcard

• Port - this can specify a single port of the protocol, if the port isn’t specificied any traffic of that protocol type is a match, An example would be just IP and no port which means all IP packets. Another example would be TCP protocol and the port 23 which would is Telnet traffic.

• Extensions - this allows check for specific flags in the IP packet. An example is is the use of the extension “established” which looks into the packet to see that it is a response to an established session. Another example is the ability to look for a specific type and code combination by the name, such as network unreachable which is type 3 code 0.

Access ListsExamples:

Access-list 2010 tcp permit any any eq 25

Access-list 1010 tcp deny 10.0.0.0 0.255.255.255 192.168.100.0 0.0.255.255 eq 80

Access-list 1100 udp permit any 10.200.100.0 0.0.31.255 eq 443

Before Midterm Material

One more example: 192.168.1.15 255.255.255.224

What are the network address, broadcast address, and host range?

Host range is 2^5 -2 (network and broadcast addresses)

32-2 = 30 host IP addresses

Broadcast address is

11000000.10101000.00000001.00011111 = 192.168.1.31

Network address is

11000000.10101000.00000001.00000000 = 192.168.1.0

Routing DecisionsRouters match the destination IP address to the network addresses in its routing table based on the most specific entry. The ordering is:

•Host address ( a /32 or individual host route)

•Subnet

•Group of Subnets

•Major Network

•Default address

If a router has 2 equal paths to the destination, it will balance traffic by sending packets through both routes. This mechanism is referred to as load balancing because it will share the load on the two paths.

Connected Routes

The first type of routes that a router inserts into its routing table are connected to its interfaces. Based on the IP addresses and subnet masks of its own interfaces, the router inserts route statements of the network addresses and the next hop as the configured interface. No additional configuration of routing or routing protocols is necessary. The administrative distance of a connected route is zero.

Connected Route Table Entry

C 10.0.0.0/8 is directly connected, FastEthernet 1/0

Static RoutesStatic routes are manually configured on a router by the network administrator. The static route has 3 parts the network IP address, the subnet mask for the network IP, and the next hop. The next hop must be either:

A specified interface on the router

An IP address already in the routing table

The interface must be up for the router to put the static route pointing to it into its routing table. If the interface goes down, the route is removed.

The next hop IP address must be in the routing table to put the static route pointing to it into its routing table. If the route table statement for the next hop address is removed, the static route is also removed from the route table.

Static Routes Cont.A static pointing to an interface routing table entry looks like:

S 172.16.0.0/12 is directly connected via FastEthernet 1/0

A static point to an IP address routing table entry looks like:

S 172.16.0.0/12 via 10.0.0.1

A static route is the only route other than directly connected that doesn’t require a routing protocol. It is used to reach a network not directly connected to the router, but reachable through one of its directly connected links.

Routing ProtocolsRouting algorithm must provide for:

•A way to pass network reachability info to other routers

•A way to receive network reachability info from other routers

•A way to determine optimal routes from the reachability info and put the best route into the routing table

•A way to react, advertise and work around network topology changes

Path Determination requires:

•Each network is connected to a router.

•Each router interface in the network must have an IP address of that network.

Routing Metrics

•Routing algorithms use metrics to determine the optimal route.

•Hop Count - the number of network devices in the path to the source

•Bandwidth - the higher the bandwidth, the lower the bandwidth network, tables or formulas to calculate the metric

•Load - metric based on the load of the links in the path, variable

•Delay - based on the amount of time the packet takes across a path

•Reliability - likelihood of link failure, variable or fixed, configurable or based on metrics

•Cost - configurable metric that allows network administrator to shift traffic as needed

DV Routing Characteristics•Periodic Updates - routing updates are sent on regular intervals with default timers that can be configured

•Neighbors - network devices that share a common data link and have the routing protocol configured on the interfaces

•Broadcast Updates - routing updates are sent via broadcast for simplicity and to make sure all necessary devices receive the updates, uninterested devices drop the update packets

•Full routing Table Updates - most often routing updates include entire routing table for simplicity, neighbors can do with the information whatever it needs to

Split Horizon•Reverse Route - a route pointing back to the router from which packets were received

•Split Horizon - technique for preventing reverse routes between two routers, for wasting resources and preventing routing loops

•Simple Split Horizon - when sending updates out an interface, do not send networks that were learned from an update that came in on the same interface

•Split Horizon with Poisoned Reverse - when sending updates out a particular interface, mark any networks that were learned from an update that received on the interface as unreachable - Considered safer and stronger

DV Routing Protocol Options•Define Infinity - to prevent routing updates from looping through a network endlessly, you can define “infinity” as a hop count, 16 is often used in distance vector routing protocols

•Triggered Updates - flash updates - update is sent by a topology or metric change immediately instead of waiting for regular update

•Holddown Timers - if a distance hop count increases, a holddown timer is set for 180 seconds for any new updates for that same route

•Asynchronous Updates - beneficial for routers sharing broadcast network preventing update packet collision - each router has own time or random time offset configured

Link State Functionality

• Each router establishes adjacency with each neighbor

• Each router sends LSAs, link state advertisements, to each neighbor. One LSA per router link, LSA includes link ID, link state, metric cost, neighbors connected to link. LSA is flooded to neighbors, which in turn floods it all its neighbors.

• Each routers stores copies of all LSAs received in a database. Databases should all be the same.

• Dijkstra algorithm is run for each destination to find optimal route, which is put into the route table.

Link State DatabaseLink state advertisements information

Router link information - router’s adjacent neighbors with Router ID, Neighbor ID, and Cost.

Stub network information - router’s directly connected networks with no other neighbors with Router

ID, Neighbor ID and Cost

Shortest Path First Algorithm• Router initializes a tree with itself as root, with cost of zero.• Cost to each neighbor is calculated and the best path is

added as a node with the lowest cost.• Each Router ID is added to the tree with the lowest cost path.• Once the tree is complete, the routing table is updated.

RIP v1 & v2

• Metric of hop count only allowable of 1-15. At 16, destination is considered unreachable, to prevent routing loops. This limits the depth of a network to run RIP.

• Update timer - Router sends gratuitous Response message out each interface every 30 seconds with full routing table.

• Expiration timer - initialized to 180 seconds for a new route and reset upon update of that route. If timer expires, hop count set to 16, unreachable, but still advertised.

• Flush timer - set to 240 seconds upon initialization, once expired, route is removed from routing table and no longer advertise.

• Holddown timer - Cisco only - set for 180 seconds when updated route has a higher hop count than previous advertisement.

RIP Version 2 Changes

• Classless routing and subnet masks in routing updates

• Routing update authentication - simple text and MD5

• Next-hop addresses for each route• External route tags• Multicast route updates, instead of broadcast• Same procedures, timers & functions of v1

OSPF TermsInterface or Link - connection from local router to attached networkLink State - status of link between two routers, relationship to neighbor routerCost - value assigned to link, based on media speedAutonomous system - group of router exchanging routing info using same routing

protocolArea - set of routers and networks that have the same area designation, each router

in an area has same infoNeighbors - two routers that share a common network, discover and exchange

routing info across itHello - Protocol used to establish and maintain neighbor relationshipsLSA - Link-State Advertisements, includes interfaces, associated cost and network

informationNeighbor Database - list of all neighbors with established two-way communicationsLink-State Database - listing of link-state entries from all other routers in area, same

database for each router in an area, generated from LSAs received

OSPF Operation1. OSPF enable router sends hello packets out all enable interfaces

• Routers with shared link may become neighbors via negotiation

2. Some neighbors form adjacencies based on network and hello type.3. Router send Link State Advertisements (LSA) over its adjacencies.

• LSA include interface link, state and cost information.

4. Router receives other LSAs and records it in its Link State Database. Then it sends the LSA out its enabled interfaces.

5. LSAs flood the OSPF area and each router has same Database.6. Router uses Shortest Path First Algorithm (SPF) to build a SPF tree with a

node for each router/destination.

7. Router uses the SPF tree to build its routing table.

*LSAs are only sent out every 30 minutes or if a topology change happens.

OSPF NeighborsOSPF routers build a neighbor table based on its exchange of hello

packets out its OSPF enabled interfacesTo become neighbors, the two routers must share a link, both interfaces

must be OSPF enable.The routers will negotiate via the use of the hello packets on the

agreement of certain parameters: Area ID, Authentication, Network Mask.

Neighbors are identified in OSPF by their Router IDThe Router ID is:• The numerically highest loopback interface IP address on the router.• If there are no loopbacks, the numerically highest physical interface

IP address of the router.

Loopbacks are preferred because of the stability of the interface ( no flapping) and the network administrator’s ability to change and assign the loopbacks.

OSPF Packet TypesHello Packet & Protocol • Used to discover neighbors• Used as keep-alives for existing neighbors• Elect Designated Routers and Backup Designated routers• 10 second interval by defaultDatabase Descriptions• Used to build adjacencies by matching LSAs in the DatabaseLink State Request• Used to request new or more recent LSAs from neighborsLink State Update• Used to flood LSAs and responds to Link State RequestsLinks State Acknowledgement• Used to ACK LSAs to make the process reliable

LSA Types1 - Router - produced by each router w/ info on interfaces, state and cost for each area it belongs to, flooded within area2 - Network - produced by each DR listing all connected routers on multiaccess networks flooded to within the area the network belongs to 3 - Network Summary - produced by ABRs to let one area know about another4 - ASBR Summary - produced by ABRs to advertise ASBRs outside of the area5 - AS External - originated by ASBRs to advertise external AS routes6 - Group Membership7 - NSAA External - allows ASBR in NSA to advertise external AS routes8 - External Attributes9 - Opaque (link local)10 - Opaque (area local)11 Opaque (AS)

OSPF Areas•Area - logical grouping of OSFP routes and links that effectively divide an OSPF domain into sub-domains.•A router in an area knows nothing detailed of topology outside of its own area.•Expressed in an Area ID of 32 bits, most often as decimal, but can be in dotted decimal like an IP address.•Area 0 must exist in all OSPF implementations and should be the backbone area of the network.•The use of areas allows administrators to cluster groups of routers together to reduce the CPU load and memory needed for running OSPF on every router.

OSPF Area Types Backbone area - only one per OSPF domain, should be area 0, receives LSA types 1 - 5

Non-backbone, Non-stub area - also receives LSA types 1-5

Stub area - receives LSA types 1-4, instead of receiving ASBR summary Type 4s with real info, default routes are inserted by ABRs to reduce routes, load and memory.

Requirements for a Stub Area:•All routers must have identical Link State Databases.•There can be no virtual links.•No router in a stub area can be a ASBR.•There can be more than 1 ABR but there is no preference.

OSPF Area Types 2Totally Stubby area - Only LSA types 1 and 2 are allowed. A special default only route version of LSA is sent by the ABRs. This reduces the routes, CPU load and memory usage even more.

Not-So-Stubby area - LSA types 1 - 4 are allowed, in addition two type 7 which allows external routes to be advertised into the area. These special Type 7 LSA are the only difference from normal stubby areas.

OSPF Router TypesInternal Router - all interfaces belong to the same area

Area Border Router (ABR) - router has interfaces in more than one area and acts a a gateway for traffic between the areas, “inter-area traffic.” An ABR must have one interface in the Backbone area. It has a separate Link State database for each area.

Backbone Router - router has at least one interface in the backbone area. Every ABR is a Backbone router, but not every backbone router is an ABR.

Autonomous Systems Boundary Router (ASBR) - a router that acts as a gateway for traffic external to the OSPF domain using routes learned through another routing protocol.