OSPF area 0 concept

[knaik99]

1.why do OSPF is structured in way that it has area 0 i.e. backbone area ? any reason.

2.if we have two LAN sites connected by MPLS VPN so Can we run OSPF AREA 1 at both LAN sites instead of OSPF area 0?

OR we have to configure OSPF configuration from AREA 0 only in above scenario?

[MHM Cisco World]

1.why do OSPF is structured in way that it has area 0 i.e. backbone area ? any reason.
You need it to interconnect other area.
2.if we have two LAN sites connected by MPLS VPN so Can we run OSPF AREA 1 at both LAN sites instead of OSPF area 0?

OR we have to configure OSPF configuration from AREA 0 only in above scenario?

Yes you can but later you may need Virtual Link, check photo below 

[MHM Cisco World]

I will share the small lab using different area than 0 connect to MPLS super backbone 

[MHM Cisco World]

this is LAB small one 
the MPLS superback area 0 connect to Area 0 to R5 and Area 0 to R6 
and area 1 to R4 
the issue since we connect R4 via area 1 then the R7 with area o can not connect to MPLS superback and advertise it prefix, 
the solution is config virtual-link between R4 and R3 (MPLS PE)

[Joseph W. Doherty]

BTW, as a side note, using virtual-links, often seems to be recommended against (i.e. don't do it).  (Sorry, don't recall "why".)

Also, you can knit together separated area zeros using tunnels.  Cannot say whether that's "better" than using OSPF virtual-links, can say, tunnels come with their own issues.

Ideally, you shouldn't need to do either.

[MHM Cisco World]

I dont link Virtual-link and not recommend use it, 
but I do lab to show @knaik99  the issue from use different area than area 0 to connect to MPLS Superback Area 0.
thanks 

[Joseph W. Doherty]

#1 OSPF's multi-area topology support is its way to provide scalability.  Area border routers can control route acceptance and/or transmission to/from the backbone area, area zero.  Even without ABR route control, area path computation, using Dijkstra’s Algorithm can be CPU resource intensive, especially as the topology it's processing increases in size.  (I.e. I recall [?] Dijkstra’s Algorithm CPU consumption increases some form of non-linearly.)  Having separate areas, bounds Dijkstra’s Algorithm's CPU impact.

#2 When using a single area OSPF topology, you can use any area number you desire.  Whether it's "better" to start with using area zero or some other number, doesn't much matter beyond the effort that might be later involved in converting to a multi-area OSPF topology.