[c-nsp] OSPF design

[Rin]

Hi all, 

Thank you all for your replies. 

I summarize some discussion points for my case:
+ For 7600 routers, it is possible to design OSPF area 0 with 100 routers
+ If we do not configure summarization on ABR router, separating the network
into different OSPF areas has no meaning in reducing LSDB size. 
+ iSPF feature cannot preventing OSPF advertise topology changes to
different OSPF areas. 
+ Deploying inter-area TE tunnels makes TE optimal path selection harder

>From these points, I am confident to configure all routers (~100) in OSPF
area 0. However, our network might be expanded in the future and more
routers will participate into OSPF. So if the recommendation of maximum 50
routers inside an OSPF area is no longer suitable for "strong" router (i.e
7600), which threshold (number of routers, number of routes, TCAM
utilization...???) should we care when design OSPF areas in ISP network?  

Thanks,
Rin

-----Original Message-----
From: Benjamin Lovell [mailto:belovell at cisco.com] 
Sent: Monday, October 25, 2010 9:22 PM
To: Rin
Cc: Heath Jones; Robert Crowe (rocrowe); cisco-nsp NSP
Subject: Re: [c-nsp] OSPF design

If you are doing MPLE TE then you really don't want more than one area as
then you get into inter-area TE tunnels which makes TE optimal path
selection harder(not possible in some cases). 

-Ben

On Oct 25, 2010, at 4:50 AM, Rin wrote:

> Dear all, 
> 
> Thank you for your replies. 
> 
> We use OSPF basically to advertise each router's loopback so that we can
> deploy L2 , L3 VPN between routers. There'll be no other external route
> advertised into OSPF. Thus, we will not configure summarization on any ABR
> router as well as stubby areas. 
> 
> I agree with Geoff's post that separating network into different OSPF
areas
> cannot reduce LSDB size. If we separate into different areas, LSA1,2,3 are
> generated and all routers must trigger SPF for a topology change inside an
> area. If we do not separate into different areas, only LSA1,2 are
generated
> and all routers must also trigger SPF for a topology change inside an
area. 
> 
> According to below statement, iSPF helps each router to run SPF only on
the
> changed portion of the topology. This means neither separating network
into
> areas nor configuring inside an area will benefit from iSPF. Correct me if
> I'm wrong at this. 
> 
> "OSPF uses Dijkstra's SPF algorithm to compute the shortest path tree
(SPT).
> During the computation of the SPT, the shortest path to each node is
> discovered. The topology tree is used to populate the routing table with
> routes to IP networks. When changes to a Type-1 or Type-2 link-state
> advertisement (LSA) occur in an area, the entire SPT is recomputed. In
many
> cases, the entire SPT need not be recomputed because most of the tree
> remains unchanged. Incremental SPF allows the system to recompute only the
> affected part of the tree. Recomputing only a portion of the tree rather
> than the entire tree results in faster OSPF convergence and saves CPU
> resources. Note that if the change to a Type-1 or Type-2 LSA occurs in the
> calculating router itself, then the full SPT is performed"  (source:
> http://www.cisco.com/en/US/docs/ios/12_0s/feature/guide/ospfispf.html) 
> 
>> From your advice, I'm more likely to configure those 100 routers inside
an
> OSPF area now.
> 
> The reason why we design OSPF up to UPE devices because we also have FTTH
> switches configure as Layer 2, also we can deploy different Layer 3
> redundancy techniques such Layer 3 loop prevention, MPLS TE..up to UPE
> layer.
> 
> Thanks,
> 
> Rin
> 
> -----Original Message-----
> From: cisco-nsp-bounces at puck.nether.net
> [mailto:cisco-nsp-bounces at puck.nether.net] On Behalf Of Heath Jones
> Sent: Saturday, October 23, 2010 6:05 AM
> To: Robert Crowe (rocrowe)
> Cc: cisco-nsp at puck.nether.net
> Subject: Re: [c-nsp] OSPF design
> 
>> Just remember
>> that you cannot summarize (today) your main Loopback used for your
>> LDP/BGP ID as there needs to be a full LSP from ingress-to-egress PE
>> across areas, if you providing L2/L3VPN services.
> 
> Is this because the lsp is label in label (outer being pe, inner being
> customer route)?
> _______________________________________________
> cisco-nsp mailing list  cisco-nsp at puck.nether.net
> https://puck.nether.net/mailman/listinfo/cisco-nsp
> archive at http://puck.nether.net/pipermail/cisco-nsp/
> 
> _______________________________________________
> cisco-nsp mailing list  cisco-nsp at puck.nether.net
> https://puck.nether.net/mailman/listinfo/cisco-nsp
> archive at http://puck.nether.net/pipermail/cisco-nsp/