OSPF(Open Shortest Path First)

1. What is OSPF?

OSPF (Open Shortest Path First) is Link State IGP Dynamic Routing protocol that operates directly over IP using protocol number 89 and AD 110.

OSPF exchanges link state information using LSA (link state advertisement) to build LSDB (link state database) which represents complete network topology.

OSPF runs SPF (shortest path first) also known as Dijkstra’s Algorithm on LSDB to calculate shortest path tree and selects best path based on metric called cost, which is derived from interface Bandwidth. Cost=reference BW(default:100Mbps)/interface BW) but can change reference bandwidth with “auto-cost reference-bandwidth xxxxxx”

OSPF uses hierarchical multi-area design with area 0 as backbone to reduce LSDB size, limit SPF recalculations, and improve scalability and convergence in large networks.

 2. What are OSPF Features?

1.Fast Convergence 

-hello and dead timers for neighbor failure detection

-Supports fast hello by using hello-multiplier command(may increase control-plane overhead)

-integrating BFD for sub-second failure detection

2.Support VLSM(variable length subnet mask) by carrying subnet mask information in LSA

3.Support hierarchical area design to reduce LSDB size, limit spf recalculation scope and improves scalability and convergence

4.Triggered Updates: Sends incremental LSAs only when topology changes and avoid full routing table updates

5.Support Authentication with Plain text and MD5 authentication.

6.Support multiple ECMP

3. What are OSPF Neighborship Requirement?

      Neighbor relationship (2-way state)

                1. Same Hello and Dead timers

                2. Same Area ID

                3. Compatible Area type

                4. Same Network (required on Broadcast and NBMA)

                5. Same Auth Type and Password

                6. Different Router-ID

       Full Adjacency (Full state)

                7. Same MTU (it is checked during DBD processing (Exstart->Exchange) and adjacency may be stuck in Exstart or Exchange. It can be bypass using “ip ospf mtu-ignore)

4. Area, Router, Network, Message and LSA Types.

Area

-hierarchical area design has two level: one is backbone area (area 0) and another is non-backbone area

-area 0 serves as backbone and transit area for inter-area traffic between non-backbone areas

-backbone area should have maximum 50 routers (design recommendation)

-one router should have maximum 3 area (design recommendation)

-reduce LSDB size and limit SPF recalculation and improve scalability and convergence

OSPF has two area categories:

                1. standard area,

                2. special area (stub, totally stub, NSSA, totally NSS).

Special area is used to reduce LSDB size and limit SPF recalculation by limiting or blocking certain LSAs.

Stub Area block type 5 (external LSAs) and ABR inject default route to stub area to reachable external network.

Totally Stubby block type 3 and type 5 and ABR inject default route to totally stubby area to reachable external network and cisco proprietary

NSSA (Not So Stubby Area)

 - blocks type 5 LSAs but allows redistribution of external routes inside NSSA as type 7 LSAs.

 - ABR translates type 7 LSAs into type 5 LSAs before advertising them to area 0 or other normal areas.

Router Types

1.Internal Router

                It has all interfaces within same area.

                Maintain a single LSDB for that area only

 

2.ABR

                -router that has interfaces in multiple areas, and one of them must be area 0

                Maintains separate LSDB per area and performs SPF calculation for each area's database.

                ABR advertises inter-area routes using Type 3 LSAs.

                Route summarization must be manually configured (area 1 range x.x.x.x x.x.x.x)

3. ASBR

                Redistribute external routes into ospf

               

                ASBR generates Type 5 LSAs (or Type 7 in NSSA)

                External route summarization must be manually configured(summary-address)

               

 

OSPF Message Types

1. Hello message

                used to discover neighbor and maintain neighbor relationship

2. DBD (Database Description)

                used to know own and neighbor's LSDB are fully synchronize or same.

3. LSR (Link state Request)

                if some LSAs are missing, request to send required LSA

4. LSU (Link State Update)

                neighbor send requested LSA

5. LSAck (Link State Acknowledge)

                -send back to neighbor when LSU message arrive

LSA Types

Type1 – Router LSA -> Every OSPF router generates its own Type 1 LSA and floods it within the same area.

                                                sh ip ospf database router #show type1 router lsa

                                                 -intra-area routes appear as “O” in routing-table after spf calculation based on type1/2 LSAs

 

                Type2 – Network LSA ->Generated by DR on multi-access broadcast and NBMA network types and flooded within same area

                                                                sh ip ospf database network

                Type3 - Summary LSA -> Generated by an ABR to advertise inter-area routes from one area into other connected areas.

                                                sh ip ospf database summary

                                                inter-area routes appear as “O IA”in routing-table

                Type4-ASBR summary LSA-> Generated by ABR and advertises the reachability to ASBR into other areas

                                                #sh ip ospf database asbr-summary

 

                Type5-External LSA -> generate by ASBR and flood to non-stub area

                                                sh ip ospf database external #show with “O E1 or O E2”

                                -(“redistribute static subnets” work specific routes only not include default route)

                -(“default-information originate” only redistribute when default have in routing-table)

                -(“default-information originate always” can redistribute without having default route in routing)

                -metric-type E1 and E2(default). E1 calculate internal cost and external cost(default 20 but for default route is 1). E2 is only external cost. E1 win then E2

 

 

Neighbor Adjacency Step by Step

---Establish Neighbor Adjacency -----

1. Down State

                -no hello received from that neighbor yet

2. Attempt State

                -specific for NBMA non-broadcast networks when neighbors are statically configured but still not received hello from neighbor

3. Init State

                -OSPF hello packets are received from neighbor but still not see its own router id from neighbor's hello message

4. Two Way state

                -Router see its own router id from neighbor's hello and it mean bidirectional confirmed. Neighbor Established

-------Synchronize OSPF Databases-

5. Ex-Start State

                - master/ slave negotiation (typically higher Router ID become master) and initial DBD sequence number negotiation, preparing to start DBD exchange

6. Exchange State

                -Start Exchange DBD by master

7. Loading State

                -router A send LSR and B reply LSU and A send back LSAck to B

 

8. Full State

                - adjacency fully formed and LSDB synchronized.