I don't think you need to do the checksumming work but just add a way to pass checksums through the index API so that when we later add the checksums, we won't have to mess with the virtualization you set up.  And, yes, each index entry which describes the location of a contiguous byte-range would have a checksum for that byte-range.  Which would be bad for example in a case of 1 byte writes since we'd have an additional 64 bytes or so in each index entry.



On Wed, Nov 13, 2013 at 12:17 PM, Chuck Cranor <chuck@ece.cmu.edu> wrote:
On Wed, Nov 13, 2013 at 01:36:13PM -0500, Bent, John wrote:
> Chuck, I know that you are doing this great project now to virtualize
> our index abstraction.  Would it be possible/straight-forward to add
> a future-enabling parameter or some other mechanism to allow a checksum
> to be passed back and forth through the index api?

So the goal is to store the checksum of all the data covered by
each index record in the index record itself?   that is doable...
the main thing i'm pondering is how to do it cleanly without
duplicate code.


DreamFactory - Open Source REST & JSON Services for HTML5 & Native Apps
OAuth, Users, Roles, SQL, NoSQL, BLOB Storage and External API Access
Free app hosting. Or install the open source package on any LAMP server.
Sign up and see examples for AngularJS, jQuery, Sencha Touch and Native!
Plfs-devel mailing list