Thursday, April 24, 2014

OSPF Network Type Broadcast, DR/ BDR Election. OSPF over NBMA. Network Type Non-Broadcast and Point-to-Multipoint

OSPF Network Type Broadcast, DR/ BDR Election. OSPF over NBMA.
Network Type Non-Broadcast and Point-to-Multipoint

Lessons Learned:

OSPF Behavior changes based on the media
EX: Ethernet vs Frame Relay vs PPP

Different media uses different “network types” to control
-how updates are sent (unicast or Multicast)
-Who forms adjacency
-How next-hop is calculated

===============================

OSPF Network Types:

-Broadcast –
-Non-Broadcast
-Point-to-point
-Point-to Multipoint
-Point-to-Multipoint Non Broadcast
-Loopback

====================================================

OSPF Network Type broadcast:

#ip ospf network broadcast
-Default on mulit-access broadcast medias
   - Ethernet, Token Ring, & FDDI
-Sends hellos and updates as multicast
-224.0.0.5 (ALL SPF Routers)  
-224.0.0.6 (ALL DRouters)
Note: Both of these multicast addresses are going to use the OSPF transport protocol 89

Performs Designated Routerd (DR) and Backup Designated Router (BDR) Elections.
DR and BDR – are used for two things.
1 – It will cut down on the number of adjacencies and LSA flooding
2 – Is for the extra recursive step in the database for building the graph of the network in side and individual area.

====================================================

DR / BDR Overview
--------------------------

- Designated Router (device in charge of receiving the LSA’a from the neighbors on the segment and then forwarding them back out to the others) (uses Multicast 224.0.0.6 for flooding). Once the DR receives the LSU in, it generates its own updates that is going back to 224.0.0.5. This is for ALL the other routers on the segment.
Uses on broadcast links to
--Minimize adjacencies
--Minimize LSA replication

-Backup Designated Router (BDR doesn’t do anything with LSA updates from neighbors, it listens for updates coming in and  Hellos coming from the DR. The BDR is used for redundancy of the DR, if it now longer hears from the DR. It will promote itself to the DR State, then there’s a new for the BDR.
--Used for redundancy of DR

Note: The DR and the BDR are listening for 224.0.0.6. All other routers are listening for 224.0.0.5.

DROther – Devices that are neither the DR and the BDR. These devices will not establish ADJ. Will Stop ant the two-way state between each other.
-All other routers link
-Forms full adjacency with DR and BDR
--Stops at 2-way ADJ with each other

DR / BDR  chosen through election process
-Election bases on two fields in the Hello packet.

Fields for elections.
Routers priority,
from 0 – 255
Higher is better
0 = never.  ( Set ip ospf priority to 0 at the link level)

If there is a tie – then use…..
Router-ID
Highest loopback / Interface IP
Can be statically set (best practice to statically set)
Higher is better.

No preemption unlike IS-IS DIS.

Note: Technically the higher Priory and higher router-id will decide the election. A device can elect itself the DR without other devices being on the network. This often depends on the boot up order.

Examples:
If I clear the ip ospf process – and debug the adjacencies – you can see the actual election process in the output.

R3#clear ip osp process
Reset ALL OSPF processes? [no]: yes


Output:

*Mar  1 00:06:58.487: OSPF: Neighbor change Event on interface FastEthernet0/0
*Mar  1 00:06:58.487: OSPF: DR/BDR election on FastEthernet0/0
*Mar  1 00:06:58.487: OSPF: Elect BDR 3.3.3.3
*Mar  1 00:06:58.487: OSPF: Elect DR 3.3.3.3
*Mar  1 00:06:58.487: OSPF: Elect BDR 0.0.0.0
*Mar  1 00:06:58.491: OSPF: Elect DR 3.3.3.3
*Mar  1 00:06:58.491:        DR: 3.3.3.3 (Id)   BDR: none
*Mar  1 00:06:58.495: OSPF: Remember old DR 192.168.23.2 (id)
*Mar  1 00:06:58.495: OSPF: 3.3.3.3 address 192.168.23.3 on FastEthernet0/0 is dead, state DOWN

*Mar  1 00:06:58.499: OSPF: Neighbor change Event on interface FastEthernet0/0
*Mar  1 00:06:58.499: OSPF: DR/BDR election on FastEthernet0/0
*Mar  1 00:06:58.499: OSPF: Elect BDR 0.0.0.0
*Mar  1 00:06:58.499: OSPF: Elect DR 0.0.0.0
*Mar  1 00:06:58.499: OSPF: Elect BDR 0.0.0.0
*Mar  1 00:06:58.499: OSPF: Elect DR 0.0.0.0     --- we do not know who the DR / BDR is.

*Mar  1 00:06:58.499:        DR: none    BDR: none
*Mar  1 00:06:58.499: OSPF: Flush network LSA immediately
*Mar  1 00:06:58.499: OSPF: Remember old DR 3.3.3.3 (id)Previously it was Router 3
*Mar  1 00:06:58.499: OSPF: Interface Serial0/0 going Down
*Mar  1 00:06:58.499: OSPF: 1.1.1.1 address 192.168.13.1 on Serial0/0 is dead, state DOWN
*Mar  1 00:06:58.499: %OSPF-5-ADJCHG: Process 1234, Nbr 1.1.1.1 on Serial0/0 from FULL to DOWN, Neighbor Down: Interface down or detached
*Mar  1 00:06:58.499: OSPF: 3.3.3.3 address 192.168.13.3 on Serial0/0 is dead, state DOWN
*Mar  1 00:06:58.499: OSPF: Interface FastEthernet0/0 going Up
*Mar  1 00:06:58.503: OSPF: Interface Serial0/0 going Up
*Mar  1 00:06:58.519: OSPF: 2 Way Communication to 1.1.1.1 on Serial0/0, state 2WAY
*Mar  1 00:06:58.519: OSPF: Send DBD to 1.1.1.1 on Serial0/0 seq 0xDFB opt 0x52 flag 0x7 len 32
*Mar  1 00:06:58.527: OSPF: 2 Way Communication to 10.45.4.4 on FastEthernet0/0, state 2WAY
*Mar  1 00:06:58.527: OSPF: Backup seen Event before WAIT timer on FastEthernet0/0  -- Now we will listen with the weight timer, to find out if anyone has already been elected. All new routers on the segment will go through this timer to make sure it doesn’t actually elect itself vs someone else. If the WAIT timer expiries and no one is elected – the router will promote itself to the DR

Note: There is  potential issue with this is -  The order that the process loads will be related to who will be elected the DR or BDR. EX: if one of the routers process loads faster than it could take over as DR. This would be true even if a device had a higher priority configured.

*Mar  1 00:06:58.531: OSPF: DR/BDR election on FastEthernet0/0
*Mar  1 00:06:58.531: OSPF: Elect BDR 10.45.4.4
*Mar  1 00:06:58.535: OSPF: Elect DR 10.45.4.4
*Mar  1 00:06:58.535:        DR: 10.45.4.4 (Id)   BDR: 10.45.4.4 (Id)
*Mar  1 00:06:58.539: OSPF: Send DBD to 10.45.4.4 on FastEthernet0/0 seq 0x12 opt 0x52 flag 0x7 len 32
*Mar  1 00:06:58.543: OSPF: 2 Way Communication to 192.168.23.2 on FastEthernet0/0, state 2WAY
*Mar  1 00:06:58.543: OSPF: Neighbor change Event on interface FastEthernet0/0

Another useful command is:
# Sh ip ospf interface (if num )

R3#sh ip ospf interface fastEthernet 0/0
FastEthernet0/0 is up, line protocol is up
  Internet Address 192.168.23.3/24, Area 0
  Process ID 1234, Router ID 3.3.3.3, Network Type BROADCAST, Cost: 1
  Transmit Delay is 1 sec, State DROTHER, Priority 1
  Designated Router (ID) 192.168.23.2, Interface address 192.168.23.2
  Backup Designated router (ID) 10.45.4.4, Interface address 192.168.23.4
  Timer intervals configured, Hello 10, Dead 40, Wait 40, Retransmit 5
    oob-resync timeout 40
    Hello due in 00:00:01
  Supports Link-local Signaling (LLS)
  Index 2/2, flood queue length 0
  Next 0x0(0)/0x0(0)
  Last flood scan length is 1, maximum is 5
  Last flood scan time is 0 msec, maximum is 0 msec
  Neighbor Count is 2, Adjacent neighbor count is 2
    Adjacent with neighbor 10.45.4.4  (Backup Designated Router)
    Adjacent with neighbor 192.168.23.2  (Designated Router)
  Suppress hello for 0 neighbor(s)
R3#

To truly eliminate this from happening you can set the ip ospf priority to zero.

Note: for Ethernet segments it doesn’t really matter who wins the Election. In ANY case where there is  only a partial mesh of Layer2 connectivity. You need to guarantee the device that has the full Layer2 connectivity – IS elected the DR.

Example: Is a hub and spoke Frame technology – you would want to make sure the HUB router is the DR.
If you have multiple Hubs one can be the DR and the other the BDR. As long as the Spoke are never elected the DR or BDR the topology will work.


OSPF Network Non-Broadcast.
Main difference between Broadcast and Non-broadcast. Broadcast is sending it’s update as multicast 224.0.0.5 and 224.0.0.6. Non-broadcast updates are sent as unicast.

#ip ospf network non-broadcast
Default on multipoint and NBMA medias.
--Frame Relay & ATM – tyoe non-broadcast.

Sends hellos as unicast
-manually defines addresses with neighbor command
-performs DR and BDR elections

This means that we need to make sure the hubs are the DR’s and we need to configure under the process the manually defined address with the “neighbor” command.

To specify the DR election we basically just tell the spokes not to become the DR or BDR. By setting the priority to 0.
Again then who is elected the DR is who’s process loads first.

Ex:
R3(config)#interface serial 0/0
R3(config-if)#ip ospf priority 0 ?

We would then need to manually configure the neighbors under the process
Note: the Neighbor commands are only required on the DR and BDR. Once the spokes hello’s come in, they will respond to originated them.

R3(config)#router ospf 1234
R3(config-router)#neighbor 192.168.13.1

Note: The default behavior of the network type broadcast or non-broadcast – the DR is not updating the next-hop value for any link state updates coming from the DROthers.


OSPF Network Type Point-to-Multipoint
#ip ospf network point-to-multipoint
-treats network as a an underlying collection of point-to-point links
-Sends hellos as multicast 224.0.0.5
Special next hop processing
Usually the best design option for partial mesh NBMA Networks

Point-to-Multipoint  - You always chose you closest layer 2 neighbor on the circuit for any routing information that is received. The Next Hop values WILL BE CHANGED.

If R1 is the Hub and it receives and Update from Spoke 1 and passed the updated to Spoke 2. The next hop value will change to R1 and not to R2.


Note: in Point-to-Multipoint  Hosts will advertise themselves as end hosts in the topology – as all /32’s

No comments:

Post a Comment