[nsp] Frame-Relay challenges

From: Vinod Anthony Joseph Cherunni (vac@dsqworld.com)
Date: Tue Sep 05 2000 - 03:38:14 EDT


Hi All,

My sincere thanks once again for the guidelines on Frame Relay networking.
I would like to take the privelege to ask a few more questions on the
same.

1. For our ISP network, We plan to have a deisgn as follows in India which
will use FR as the medium.
 
Tier 1 cities - Chennai Mumbai Delhi (Internet exit points, connected to
each other using a Full FR mesh)

Tier 2 cities - (a) Bangalore - Which will connect to Mumbai & Chennai
                        (b) Calcutta - Which will connect to Mumbai &
Delhi
                        (c) Ernakulam - Which will connect to Mumbai &
Chennai
                        (d) Hyderabad- Which will connect to Mumbai &
Chennai

Tier 3 cities - 30 locations which will connect to the nearest Tier 1, &
Tier 2 city forming a regional backbone.

Tier 1 & Tier 2 cities will form the OSPF core (Are 0). Tier 3 cities will
hang off the nearest Tier 1/ Tier 2 city as mentioned above, & will form
seperate areas along with the parent Tier 1/ Tier 2 city which will be
totally stubby areas.

I felt that the Tier 1 & Tier 2 cities will be interconnected to each
other using Frame Relay point-point interfaces, since the connection is a
partial mesh & not in a hub & spoke fashion.

Also I feel that the connection between the Tier 3 cities & Tier 2 cities
could use multipoint interfaces. My concern is that there could be a
probability that some critical Tier 3 cities will have a primary
connection to one of the Tier1 / Tier 2 cities & secondary connection to
a another Tier 1 / Tier 2 city. I will ensure that such as a situation
will not result in a misconfigured OSPF design, because I would ensure
that a city will not span OSPF areas. Will this effect the multipoint
configuration, because this will not be a true hub & spoke config.

Also I see mentions in the Cisco website stating that Point-Point config
is the way to go. I have observed that the real advantage in multipoint
configurations is IP address conservation, Suppose I allocate a Class C
for addressing my WAN links, In a multipoint config I could use all 254
addresses, But in point-point config I will end using only half (64
subnets), But I can probably compensate using IP unnumbered. But apart
from that will my routing table show 64 subnets seperately when I use
point-point configuration, against 1 class C network being shown if used
with Multipoint configurations.

Kindly advice

Thanks & Warm regards,
Vinod.



This archive was generated by hypermail 2b29 : Sun Aug 04 2002 - 04:12:16 EDT