One point I would like to confirm from the current draft is the separation of 
the routing key from the storage and retrieval key. By routing key I mean the 
resource id/node id in the destination list and storage/retrieval key I mean 
the resource id in the storereq or fetchreq. There are several important 
reasons why we may want to store/fetch a data value with resource id X on a 
node with node id Y where Y is not the owner of X. This can be considered as a 
targeted STORE than a routed STORE.

This is important for the following immediate reasons:
1. Successor replication which is important to have for data availability. Some 
of the extensions such as the self tuning draft require successor replication 
(a targeted STORE of resource id X on node id Y where Y is not the owner of X)
2. Caching is likely to be an important part of a successful p2p system and 
this enables targeted STOREs to be performed to nodes that will cache 
information.

Just wanted to confirm that this is intended as part of the draft.

Thanks
Saumitra
_______________________________________________
P2PSIP mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/p2psip

Reply via email to