الفهرس Only 14 pages are availabe for public view
 |  | from | 125 |  |  |
| | | from | 125 | | |
Abstract
Object relational database management systems (ORDBMSs) are becoming more popular in storing anc ’eving XML data than native XML DBMSs. In most ORDBMSs, an XML document is stored as Characte Object (CLOB) inside the relation. Efficient XML parsers are used to retrieve the desired query nodes cr, less attention is given to XML update queries. With the upcoming standardization of XML updatl
’es, the current implementation of lock granularity in database systems imposes a great limitation on thl ncy of parallel transactions. This motivated several experimental ORDBMSs to shred the XML node: internal relations. However, this approach has its drawbacks.
This thesis proposes an adaptive technique for selective shredding. It is based on an existinj lementation and takes the changes in the workload pattern into consideration. In the proposed approach XML document is gradually shredded into smaller XML portions to achieve higher concurrency fo cry updates. The XML Access Monitor logs the activities of XQuery transactions for each XMI ent stored in the database. After a while, the XML Access Monitor checks these activities and identifie:
XPathes that are frequently updated for each XML document. The XML document is split at the node tha
being frequently updated. The XML sub-trees that are frequently accessed will be stored separately bJ torage Manager which helps to achieve higher concurrency of access on these sub-trees. When the access 01 shredded XML portions decreases, the Storage Manager consolidates these portions back into a greate document\portion.
The performance of the proposed approach is analyzed and compared to the CLOB and the complete ding approaches. The conducted performance study showed that the proposed technique performs vef’) II, The response times, transaction aborts, throughput, and space consumption are reasonable. Since the
approach adapts to the changes in the workload, it can be useful for many applications.