Just to add a little more on this topic, we expect this will be the really interesting part of migration, optimizing around I/O requests made to a file. Here are a couple of scenarios to consider:
If a request is small on a large file, DMM could return the portion requested to minimize the latency exposed to the waiting thread. However, this needs to be done with some care. In certain file systems, a contiguous file could be fragmented by this action. Requests that make up larger percentages of files will likely cause the entire file to be migrated. Repeated access to attributes of files could cause DMM to pre-fetch the file in anticipation of an eventual data access. Upshot is there are many ideas that we can consider in the case where a migration is triggered by a client's access to a file. Mark This message posted from opensolaris.org _______________________________________________ storage-discuss mailing list [email protected] http://mail.opensolaris.org/mailman/listinfo/storage-discuss
