OSPF area - migration to new address scheme & area type

[iores]

Hi,

I have OSPF area 1 that I would like to migrate to a new address scheme for P2P links, and new area type - Totally Stub. Currently, only SVI interfaces are used (no routed ports).

I was thinking of a way to minimize downtime. My idea is to create a new "parallel" network with new P2P links and SVI interfaces. These new interfaces and advertised networks would be part of the newly created OSPF area 2 (the current is area 1). Then, when everything in area 2 would be up and running, I would shutdown everything from area 1.

Is this approach OK?

[MHM Cisco World]

check below

MHM

[iores]

Hi,

I will test it in lab, definitely. Just wanted to validate the theory first with the community.

What do you mean with 'max-metric'?

[MHM Cisco World]

Check below

MHM

[iores]

Thanks for the link. What about approach I proposed - do you find it OK generally or is there a better way?

[paul driver]

Hello
FYI -Max-metric lsa depends on your topology it maybe you won’t be able to implement as it maxes out that particular ospf router process cost so you will require at least two router processes if not you will just max out the path to the other router(s)

I would say most simplistic solution would be to create additional links and cost out them out, or even just add secondary addressing and at a time applicable make those secondaries -primary's as such the old adjacency will be torn down and new one established  but again this is dependent on your topology and what you are limited to do

Can you share a topology diagram and maybe we can provide a more definitive option for you?

[Joseph W. Doherty]

I agree with Paul, more info would be useful.

BTW, as you mention p2p and not currently having any routed ports, you do realize an OSPF p2p doesn't require either a routed port or a /30 or /31?

No edge network addressing charges?  I.e. just infrastructure addressing changes?

Doing a parallel area should minimize downtime, but there will be a hit when you migraine edge networks.  Also, unless you have redundant paths, there can be hits to transit traffic too, and even with redundancy, you may temporarily lose that too.

How "big" is the area?

I've been part of a couple, rather large, OSPF area changes.  All done during a planned maintenance.  Cannot say that's a better approach, much depends on your network needs.  I believe a parallel approach may reduce downtime for hosts but also makes it easier to introduce subtle routing errors.

[MHM Cisco World]

Thanks for all reply'

If he have area1 and advertise prefix into area1 then additional link in area2 not solve issue' 

Still the ABR advertise prefix in area1 one only.

That come from interface can not be config in two different areas.

So he have to change the prefix from lsa1/2 into lsa5 (redistrubte connect) this way both areas can advertise prefix' but here we face issue of downtime.

Thanks and happy new year to ALL

MHM

[Richard Burts]

It seems to me that in doing the implementation you will need to start with the device(s) that act as Border Router. Configure interface(s) to change from area 1 to area 2, and change the neighbors on those links. That way area 1 and area 2 have access to the backbone area. Then go to the neighbor routers and change their areas.

[paul driver]

Hello

---

@MHM Cisco World wrote:

That come from interface can not be config in two different areas.

---

Apologies however your statement is incorrect, you can have a interface in multiple ospf areas which can be most applicable for path selection but again has ive stated it all depends on the topology  

[MHM Cisco World]

router ospf xxx 

network <subnet of interface> area 1

network <subnet of interface> area 2 <<- this override the area 1 and hence you need to config subnet as LSA5 as I mention above

MHM

[paul driver]

Hello @MHM Cisco World 
Ospf allows an interface to be in multiple areas, and this is done via the interface not a router process network statement and redistribution is NOT required

Example:
rtr1
router ospf 1
router-id 1.1.1.1
int x/x
ip address 1.1.1.1 255.255.255.252
ip ospf network point-to-point
ip ospf 1 area 0
iip ospf multi-area 1
ip ospf multi-area 2
etc...


rtr2
router ospf 1
router-id 2.2.2.2
int x/x
ip address 1.1.1.2 255.255.255.252
ip ospf network point-to-point
ip ospf 1 area 0
iip ospf multi-area 1
ip ospf multi-area 2
etc...

[MHM Cisco World]

You still not get my idea

CCheck this topolgy 

MHM

[paul driver]

Hello
So in your topology the LO prefix will always take its own intra-area path first be it  area1 or 2 , so why not have BOTH interfaces in the same area and cost out the less preferred one.... Much simpler dont you think?

[Joseph W. Doherty]

BTW, the OSPF multi-area could be a fine approach, but OP mentioned he also wanted to readdress and wants minimal downtime.

If he readdresses first, without redundancy, he'll likely cause a temporary traffic hit, and even with redundancy, he'll have a temporary loss of redundancy.  Of course, such traffic hits, assuming done correctly, will be very short lived, although done correctly, but poorly, even redundancy can cause a brief hit during re-convergence.

(To minimize a redundancy re-convergence hit, you should insure backup path appears as the better path before taking the original better path out.  This is one problem with the max-metric approach.  [Mind you, again, usually, finding the better path is pretty quick, but redundant path re-convergence can drop or delay packets, if done using the "bad news" approach vs the "good news" approach.  If any are unfamiliar with the difference, let me know and I'll explain further.  Again, most of the time, the "good vs. bad news" approach doesn't matter enough we need to concern ourselves, but when trying to avoid any needless network hit, it's worth understanding.  It can become important in certain situations, like especially sensitive loss/delay traffic, e.g. VoIP, across intercontinental WAN links.])