×

Warning message

  • Cisco Support Forums is in Read Only mode while the site is being migrated.
  • Cisco Support Forums is in Read Only mode while the site is being migrated.

Routed access OSPF design

Endorsed Question
Nov 29th, 2013
User Badges:

Hi All,


I need some advise on the following design.


I am designing a campus network using routed access with OSPF. I will have 3 tiers - core, distribution and access. I am looking to configure the core/distribution links in OSPF area 0 and then the distribution/access links in a non-backbone area such as 1. I will then be able to filter/summarize etc as the ABR boundary.


I only have 2 links from my core to distribution layer and I'm planning to terminate them on separate distribution switches (dist1 and dist2), see attached diagram.


The links facing the DC will be in area0 and the links facing the access will be area1 but what area should the distribution to distribution link be in? I understand that if it is in the non-backbone area1 then I will loose the ability to use the link to the second core as intra area routes will be preferred by dist1. Are there any other considerations here? Should I connect two links between the the 4500s and configure 1 link in area 0 and 1 link in area1?


Thanks,

            

Endorsed by Daniel Bruhn
daniel.dib about 3 years 8 months ago

I would add a link between Dist1 to Core2 and Dist2 to Core1 if possible. This will improve convergence as you would have triangles instead of square topology. You would then have equal cost routes towards the access switches so if a link fails then there is already a route in the RIB/FIB.


Regarding which area to put the Dist to Dist link in it would depend on if you expect most traffic to arrive towards area 0 or toward other areas. By the look of your design it seems you will be using mostly the backbone to transport traffic and put hosts/servers etc in their own areas?


Another option is to have dual adjacencies between the Dists. One in area 0 and one in area 1. Then you would have optimal forwarding even if a link fails.



Daniel Dib
CCIE #37149

  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 0 (0 ratings)
Loading.
Jon Marshall Fri, 11/29/2013 - 03:39
User Badges:
  • Super Blue, 32500 points or more
  • Hall of Fame,

    Founding Member

  • Cisco Designated VIP,

    2017 LAN, WAN

You need to make the link area 0. If the link is area 1 and then the link between the cores failed then you have a partitioned area 0 and you don't want that.


And as you say the 2 dist switches will be seeing area 0 routes as inter area routes via each other if you put it in area 1.


I can't see an advantage to having another link in area 1 between the 2 switches. The dist switches each have a link to each access switch so you have redundancy there anyway.  Assuming the link would be a routed link you don't gain anything anyway because if access1 to dist1 failed traffic would be sent to dist2. Dist2 would see a better path via core2 because the path to the core via dist1 is an extra hop away.


Perhaps others can offer advice as well but personally i would go with area 0 for that link and i can't see the need for another link in area 1.


Jon

Jon Marshall Fri, 11/29/2013 - 03:52
User Badges:
  • Super Blue, 32500 points or more
  • Hall of Fame,

    Founding Member

  • Cisco Designated VIP,

    2017 LAN, WAN

I am assuming that all links shown in the diagram are L3 routed links and not L2 trunks.


I would still be very interested to hear others thoughts on this regarding an extra link for area 1.


Jon

daniel.dib Fri, 11/29/2013 - 04:45
User Badges:
  • Silver, 250 points or more

I would add a link between Dist1 to Core2 and Dist2 to Core1 if possible. This will improve convergence as you would have triangles instead of square topology. You would then have equal cost routes towards the access switches so if a link fails then there is already a route in the RIB/FIB.


Regarding which area to put the Dist to Dist link in it would depend on if you expect most traffic to arrive towards area 0 or toward other areas. By the look of your design it seems you will be using mostly the backbone to transport traffic and put hosts/servers etc in their own areas?


Another option is to have dual adjacencies between the Dists. One in area 0 and one in area 1. Then you would have optimal forwarding even if a link fails.



Daniel Dib
CCIE #37149

willscotty2012 Fri, 11/29/2013 - 06:32
User Badges:

Hi,


Yes, all links will be L3.


Unfortunately I don't have the luxury of extra links otherwise I would follow your recommendations and connect each dist switch to both cores, I'm stuck with the two links :-(


Area0 will be our data centers so this will contain servers and external connectivity such as Internet and MPLS. Non backbone areas will mainly be hosts.


Also there could be a possibility that hosts are connected directly to Dist1 and Dist2. If these hosts were connected to subnets in Area1 then they might use access1 as transit if I don't have a Dist to Dist connection in area1.


In the full design there will be a large number of access switches and VLANs with  /24 networks so I want the ability to summerize these networks up to the core.

daniel.dib Sat, 11/30/2013 - 00:47
User Badges:
  • Silver, 250 points or more

Yes, summarization is always good.


Seems the majority of the load will be in area 0 so first priority would be to have a link in area 0 between the Dist switches.


Like you said though the access switches could end up as transit. This is up to you if it will be a problem. Are the Dist to Dist links larger than the Access to Dist links? What you could do then is like I described to put two adjacencies between the Dists. However you would need an adjacency for every area that is present on the Dist switches so it depends on the number of areas if this is a reasonable solution.


Remember that an ABR will have a LSDB for every area and needs to run SPF for each DB. It also must flood type 3/4/5 LSAs to these areas, maintain adjacencies and so on. On modern equipment this should not be too tough to handle but it all depends on the scale.



Daniel Dib
CCIE #37149

Joseph W. Doherty Sat, 11/30/2013 - 03:51
User Badges:
  • Super Bronze, 10000 points or more

Disclaimer


The  Author of this posting offers the information contained within this  posting without consideration and with the reader's understanding that  there's no implied or expressed suitability or fitness for any purpose.  Information provided is for informational purposes only and should not  be construed as rendering professional advice of any kind. Usage of this  posting's information is solely at reader's own risk.


Liability Disclaimer


In  no event shall Author be liable for any damages whatsoever (including,  without limitation, damages for loss of use, data or profit) arising out  of the use or inability to use the posting's information even if Author  has been advised of the possibility of such damage.


Posting


About how many L3 routing devices, and how many subnets, do you anticipate?  Depending on that, you might one to consider a single area design.

Actions

This Discussion