Anil
You are correct that graceful restart was originally designed for platforms with redundant sups or redundant processors (like the ASR) and makes good sense on these platforms since the data plane would be capable of continuing to forward traffic while the control plane establishes/re-establishes the routing protocol information.
But the capability of graceful restart has been extended to the ISR platform, which does not have the redundant sup or processor. In this case the benefit of graceful restart is not in supporting continued forwarding of traffic. But the advantage is in preventing the flap that would occur when the neighbor takes down the neighbor relationship, removes all the routes (and advertises the loss of route to its neighbors), re-establishes the neighbor relationship, re-learns the routes (and advertises the re-learned routes to its neighbors). With graceful restart the ISR can signal to its neighbors that it is restarting. The neighbor marks the routes as stale but they remain its its routing table. Assuming that the restart completes successfully within the "grace period" provided by graceful restart, there is no flap, no purging of routes and re-learning of routes, and no advertisement to neighbors about route changes. This provides greater stability in the BGP network and that can be quite helpful.
The Cisco documentation for 15 is specific that graceful restart is supported on the ISR routers and I believe that this is why it makes sense. I believe that you would have no problem and could have advantages in configuring graceful restart as your client requests.
You might look at this link for additional details:
http://www.cisco.com/en/US/docs/ios/iproute_bgp/configuration/guide/irg_adv_features.html#wp1056277
HTH
Rick
HTH
Rick
