Information Technology Reference
In-Depth Information
include storing multiple copies of the data at separate physical locations, employ-
ing data protection mechanisms such as RAID, performing periodic media refresh,
etc. However, LTDP systems will be more robust and have less probability for data
corruption or loss if their storage component supports also logical preservation. We
call such storage components preservation-aware storage.
Preservation DataStores (PDS) are OAIS-based preservation-aware storage
[
186
,
187
] that focuses on supporting logical preservation in addition to the tra-
ditional bit preservation. PDS is aware of the structure of an archival information
package (AIP), and offloads functions traditionally performed by applications to the
storage layer. These functions include handling AIP “metadata”, calculating and val-
idating fixity, supporting authenticity processes, managing the AIP representation
information (RepInfo) and validating referential integrity. A unique and innovative
capability of PDS is the support for computation near the data; a paradigm that
moves the execution module to the location of the data instead of moving the data to
the execution module's location. To achieve this, PDS enables the load and execu-
tion of storlets, which are execution modules for performing data intensive functions
(e.g., data transformation) close to the data. This saves network traffic and improves
performance and robustness. Additionally, this enables optimal scheduling of tasks
(e.g., performing data transformation during bit migration saves repeated reading of
massive amounts of data).
Tape storage systems and disk storage systems are currently the prominent types
of media on which data is preserved. In many cases, the preservation data tends to be
cold (inactive) and is seldom accessed over time. Tapes are attractive in these cases
as they are more reliable than disks and their expected lifetime is 3-10 times higher
than that of disks. Additionally, tapes consume 25 times less power than disks. Thus,
overall, tapes are much more cost-effective than disks and are especially attractive
for preservation. PDS is flexible, able to use any type of media as well as able
to be used for any type of data. It supports placement of the AIPs in containers
where each such container is self-describing and self-contained. This capability is
especially useful for offline storage media.
PDS serves as the infrastructure storage of CASPAR and was installed and inte-
grated at Europe Space Agency (ESA) where it was tested with scientific data. PDS
is integrated in CASPAR graphical user interface and can be used directly or via
the PACK component that packages raw data into AIPs and calls PDS to store
them. PDS implements and supports the CASPAR OAIS-compliant authenticity
model that includes authenticity protocols and steps. PDS interfaces are published
in SourceForge. Finally, PDS is available for public download and free evaluation
at alphaWorks [
188
].
17.6.2 PDS Description
In this section we describe PDS architecture, its detailed functionality and the means
to ensure this functionality and to extend PDS over time.
