Hi
Before you get onto design something inline with your query would suggest to classify the kinda traffic patterns in your network.
Read and observe the B/W requirements and the nature of them like whether they are greedy in nature or smooth or bursty..
Once you have these points in paper you can start preparing the next stage in assigning the reqd bandwith to these applications or the kinda priority which is reqd for the traffic which is latency sensitive or drop sensitive (voice).
Like as you have mentioned FTP here its bulk in nature and depends upon the requirement for data transfer frequency which is again proportional to the business needs and demands.
So as you mentioned in your first point to always have the reservation for FTP i dont think thts the best practice being carried out all throughout the network coz in tht case u will end up in B/W scarcity for other applications
My suggestion would be configure 1Mbps B/W for FTP traffic and if you have some other applications allocate the B/W reqd for them based on the nature or the traffic and the kinda latency it requires to work properly.
so in the first case you can restrict all the other traffic using rate-limit upto the aval b/w and keeping the FTP to make use of full 1Mbps b/w in the E1 pipe.
Regarding your second point as already mentioned create different classess and bind them in ur policy map and allocate the bandwidth under them accordingly as per your requirement.
Also do remember that these qos policies have to be identical all throughout the network thts from the source to destination to have real effect and to have desired/expected results.
Otherwise the bandwidth gurantee may break in between in traversal path which wont help u the kinda expected results .
regds
