Hi Margaret,
>
>>For OSPFv3, as you described in I-D, we are allocating separate OSPF
>>process for each area. The current OSPFv3 does not consider SBR at all.
>>This is the reason for separating OSPF processes.
>
>Do these processes share a single global routing table, based on
>the link-state advertisement from all peers? Or perhaps you run one
>"global" process and one "site-local" process per site?
Currently, processes for sites are completely separated.
Let's say, a router has 4 interfaces, if1, if2, if3, and if4.
Sites are configured as:
Site1 has if1 and if4.
Site2 has if2 and if3.
In our current implementation, two OSPF processes are created as:
OSPF process 1 runs for Site1.
OSPF process 2 runs for Site2.
On each process, all routing information, including global and
site-local are handled. Because of this, a user need to "redistribute"
global prefixes (not site-local prefixes) each other.
This is the demerit of separating OSPF process per site.
As you described in the previous mail, we could run one "global" process
and one "site-local" process per site. In this case, a interface need to
belong to two OSPF processes, one for global and another for site.
In this case, "redistribution" of global prefixes among sites will not
be required.
In both cases, multiple OSPF processes are required since global and
site-local information could appear in single LSA according to
the current OSPFv3 RFC.
if1
| /Site Boundary
| /
| /
+---------+-----------+
| /|
| Site1 / |
| / |
if4-----------+ --------------- +---------- if2
| / |
| / Site2 |
|/ |
+----------+----------+
/ |
/ |
/ |
/ |
if3
Thank you,
Hiroki Ishibashi
[EMAIL PROTECTED]
--------------------------------------------------------------------
IETF IPng Working Group Mailing List
IPng Home Page: http://playground.sun.com/ipng
FTP archive: ftp://playground.sun.com/pub/ipng
Direct all administrative requests to [EMAIL PROTECTED]
--------------------------------------------------------------------
