Databases Reference
In-Depth Information
record anonymity provided by keyword merging critically depends on the set
of keywords that are merged together, and this issue needs further investiga-
tion.
9 Open Problems
In this section, we briefly summarize the biggest open issues and challenges
in trustworthy records retention.
•
Corrections
. Current models for records retention do not support cor-
rections to record content. However, corrections are often necessary in
practice (for example, when medical information is placed in the wrong
record for a patient with a common name), and are mandated by law in
some domains. An elegant, cost-effective approach is needed for supporting
corrections and providing trustworthy provenance information and audit
trails.
•
Deletions
. As discussed above, no entirely satisfactory scheme exists for
trustworthy deletion of records. Traces of record metadata may remain in
indexes or migration logs, potentially allowing an adversary to infer the
contents of a deleted record.
•
Structured information
. This chapter has focused on the trustworthy
retention of unstructured or semistructured data. Database records need a
similar level of protection, but no work to date has addressed this problem.
•
Exploiting trusted hardware
. In addition to the trusted hardware
based WORM layer, we believe it is important to explore how to deploy
such hardware to achieve increased security and eciency in the upper
(e.g., indexing) layers. The challenge is to exploit a very small, cheap
piece of trusted hardware along with off-the-shelf regular storage system
components, in an architecture that provides good performance.
In addition to these major problems, many smaller questions remain open.
For example, there are no trustworthy approaches to multidimensional index-
ing.
Acknowledgements
Authors Ragib Hasan, Soumyadeb Mitra, and Marianne Winslett were sup-
ported by NSF awards IIS-0331707, CNS-0331690, and CNS-0524695. Author
Radu Sion was supported in part by the Stony Brook Oce of the Vice Pres-
ident for Research and by the NSF award CNS-0627554.
