Solved: MPLS PE CE Route Summarizatio

snarayanaraju · ‎01-12-2010

Dear Friends,

This is about MPLS PE-CE OSPF as connected below

CE1----------------PE1------P-----PE2--------------CE2

| |

area 1 area 2

I have multiple prefix that are advertised from CE1 to CE2. They are

172.16.0.1/24

172.16.1.1/24

172.16.2.1/24

172.16.3.1/24

To sumarize these prefix I am using the command area 1 range 172.16.0.0 255.255.252.0 in PE2 Router under ospf vrf process. But it is not working.

Then I tried with summary-address 172.16.0.0 255.255.252.0 then it is working as expected, But..... the route (172.16.0.0/22) is now advertised to CE2 as E2 type.

I also got the clue that PE2 & PE1 Routersuingare treated as ASBR as it is redistrib

My concern is, whether this is a default behaviour or I am doing something wrong ??

Thanks in advance

sairam

Giuseppe Larosa · ‎01-12-2010

Hello Sairam,

I've never tried this as I said however, the following considerations should apply:

if there wasn't the MPLS L3 VPN where would you put the area range command, given that PE1 and PE2 would be backbone routers and CE1 and CE2 two ABRs CE1 ABR (1,0) and CE2 ABR (0,2) ?

area range can be used at area boundary on ABR.

I think your chances to use it with success increase if you use it on CE1 before injecting the routes on area 0 at CE1-PE1 link.

There is no sense in making component routes to travel into MP BGP and then to attempt to summarize them at remote PE PE2.

Try to use the command on CE1 and observe results on PE2 and CE2.

Edit:

if PE1-CE1 link is in area 1 the ABR is PE1 and you should use area range on PE1

Again I would expect this to be more effective then trying to apply the command on PE2 for routes advertised by CE1 ot PE1 in area 1.

Hope to help

Giuseppe

View solution in original post

Giuseppe Larosa · ‎01-12-2010

Hello Sairam,

when using OSPF as PE-CE protocol the OSPF process-id should match on local and remote PE to have routes imported as O IA.

If different OSPF routes are created but as external routes.

This was my finding during my tests some years ago.

The reason is that the OSPF process-id is carried within a BGP extended community.

Are you using the same or different OSPF process-ids on PE nodes?

Also area-id should be area 0 on both PE-CE links

Hope to help

Giuseppe

snarayanaraju · ‎01-12-2010

Hi Giuseppe,

Thanks. Yes Both the sides I am using the same process ID. I am aware that if process ID differs the rotues will become external when adverised from PE to PE.

But the concern here is, when the routes are Internal (I OA), where to apply the "AREA x RANGE x.y.z.w" command. Whether it has to be applied in PE which is showing as ASBR??

I tried various combinations.Only summary-address is working where as area range is useless

thanks in advance

sairam

Giuseppe Larosa · ‎01-12-2010

Hello Sairam,

I've never tried this as I said however, the following considerations should apply:

if there wasn't the MPLS L3 VPN where would you put the area range command, given that PE1 and PE2 would be backbone routers and CE1 and CE2 two ABRs CE1 ABR (1,0) and CE2 ABR (0,2) ?

area range can be used at area boundary on ABR.

I think your chances to use it with success increase if you use it on CE1 before injecting the routes on area 0 at CE1-PE1 link.

There is no sense in making component routes to travel into MP BGP and then to attempt to summarize them at remote PE PE2.

Try to use the command on CE1 and observe results on PE2 and CE2.

Edit:

if PE1-CE1 link is in area 1 the ABR is PE1 and you should use area range on PE1

Again I would expect this to be more effective then trying to apply the command on PE2 for routes advertised by CE1 ot PE1 in area 1.

Hope to help

Giuseppe