Routing

Hi, experts i have attached OSPF design for a production network , Designer has implemented only a single area on all links (links connecting Access Router to Core Router and link connecting Access link to another Access Router) . It was hard to me to under stand about design intention , specially placing two link of an access link in same OSPF area (one link connecting to Core Router and Other link connecting to another Access link)

 

Need your expert opinion about pros and cons of this design

Hello,

Pretty sure this is one of those 'philosophical' discussions. My personal rule is to have different OSPF areas only on a 'need-to' basis.

With that said, you always need to plan out your OSPF design with these factors in mind:

1. How would the network scale in the long-term?
2. How would the network scale affect the hardware that we're using?
3. Do we need summarization, redistribution (NSSA etc)?

If your routers can take the load of a single area (relatively bigger OSPF database) in the long-term, then I would stick with the single area approach. But that's just my personal opinion. 🙂

HTH,
ankit

Ok to some extent single area is not an issue because network is too small, but my actual question is that placing access link between two access router (branch devices) is a wise approach. What will be its effect , positive or negative

Hello,

You asked for it 🙂

1/ This design is suboptimal because the OSPF initial area number chosen ot be 0.0.0.12, not 0.0.0.0.

If only 1 extra OSPF area need to be added, this will still work fine and Jeff Doyle's book on OSPF and ISIS has an example of 2 nonzero OSPF areas talking to each other.

But imagine what happens if 2 or more extra nonzero OSPF areas need to be added.

With 2 or more extra nonzero OSPF areas need ot be added, then they cannot be arranged in a straight line:

0.0.0.12--0.0.0.13--0.0.0.14. In this fashion 0.0.0.12 cannot talk to 0.0.0.14 and vice versa.

2/ You are asking whether placing a link between 2 access routers in same 0.0.0.12 area is correct.

If this link is placed in a separate nonzero area, then it is clearly incorrect, because this link will never be used. The reason can be found in RFC 2328 : Type-3 LSA received over nonbackbone interfaces are ignored.

3/ I hope the lnks are properly sized because if one Core Router-Access Router link fails, then the remaining Core-Access link has to carry double or even triple usual amount of traffic.

HTH

Thanks

Alex

there is no issue as far as the design is concerned . all the links are in the same area 12 LSA will be exchanged by DR. however DR if selction will decide the LSA exchange phenomenon. seems no problem

Ok , type 2 LSA (Network LSA) will be advertised by selected DR in Segment , now consider following situation

 

 Please view attachment before reading next Para

 

Here Core router is advertising an external route through an export policy , that route will reach to each Access Router through type 4 LSA via directly connected links (Access to Core) and route  is installed in routing table of both routers. Now both router will have a directly connected link configured in same OSPF area, both router will also share the LSA received from Core router. Now my question is which route version will be installed in routing table of access routers as they both are receiving same route from Core router and also sharing same route between each other, Is not going to create a routing loop , Please correct me if i am wrong

Hi there,

 

---

@Route-Champ wrote:

 

Here Core router is advertising an external route through an export policy , that route will reach to each Access Router through type 4 LSA via directly connected links (Access to Core) 

---

Incorrect. Type-4 LSA is used for ASBR reachability.

 

 

---

@Route-Champ wrote:

 both router will also share the LSA received from Core router. Now my question is which route version will be installed in routing table of access routers as they both are receiving same route from Core router and also sharing same route between each other, Is not going to create a routing loop , 

---

The route with least metric will get installed. How external route metric is calculated - please see RFC2328 section 2.3

http://www.ietf.org/rfc/rfc2328.txt

 

HTH

Thanks
Alex

Let me clarify on first part , Core router is advertising an external route into AS , can we not call it ASBR , if it is ASBR  then should not advertise type 4 LSA into AS ..... Sorry to be too naive .....

 

 

Thanks for posting rfc link i will defiantly go through for better understanding of OSPF

Hello,

 

---

@Route-Champ wrote:

Let me clarify on first part , Core router is advertising an external route into AS , can we not call it ASBR , if it is ASBR  then should not advertise type 4 LSA into AS ..... 

---

If a OSPF-speaking router advertises an "external" route into OSPF, it is called "AS boundary router" in RFC 2328.

You have no other choice  if You want to be RFC 2328- compliant 

 

And ASBR does not advertise Type-4 for self, it is done by ABR.

 

HTH

Thanks
Alex

 

 

Thank a lot aarseniev , now i am pretty much clear about OSPF external route selection mechanism after going through RFC2328 section 2.3  and reading material available at link http://www.juniper.net/techpubs/en_US/junos13.1/topics/concept/OSPF-routing-external-metrics-overview.html. By default external route are advertised as type 2 external metric (in Juniper) and  if two equal cost external route reached to a router  then OSPF uses internal link cost (to external route advertising routers) as tie breaker for external route next hop selection mechanism

 

 

 

 

 

Pleas mark this as accepted solution if it solves your problem

Kudos a good way of appreciation

 

Kashif Nawaz

JNCIP-Sec , JNCIS-Ent, JNCIS-Sec, JNCIA-Junos