You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
No fragment should be duplicated in the filesystem to a greater extent than demanded by the most redundant storage scheme using it. In other words, fragments should be reusable across schemes and clients.
For example, if two different files are stored that happen to share a shard, and if one file is stored so that each shard is duplicated twice, while the other is stored so that each shard is duplicated thrice, then there should be only three copies of the shared shard stored in the filesystem, not five.
If a client requests to delete a file from the filesystem, then fragments should only be deleted if there are no other pointers to them from other files stored by other clients. Ideally, if the client happens to be employing the most redundant scheme affecting a particular stored fragment that has other clients, then the redundancy of that fragment should be reduced to the next most redundant scheme, although if determining this is hard it is not that important to do since we are still assured that "at least as much redundancy" exists in that shard as is required by any client.
The text was updated successfully, but these errors were encountered:
No fragment should be duplicated in the filesystem to a greater extent than demanded by the most redundant storage scheme using it. In other words, fragments should be reusable across schemes and clients.
For example, if two different files are stored that happen to share a shard, and if one file is stored so that each shard is duplicated twice, while the other is stored so that each shard is duplicated thrice, then there should be only three copies of the shared shard stored in the filesystem, not five.
If a client requests to delete a file from the filesystem, then fragments should only be deleted if there are no other pointers to them from other files stored by other clients. Ideally, if the client happens to be employing the most redundant scheme affecting a particular stored fragment that has other clients, then the redundancy of that fragment should be reduced to the next most redundant scheme, although if determining this is hard it is not that important to do since we are still assured that "at least as much redundancy" exists in that shard as is required by any client.
The text was updated successfully, but these errors were encountered: