Managing Multiversion Documents

1
Managing Multiversion Documents Historical
Databases a Unified Solution Based on XML

• Fusheng Wang
• Siemens Corporate Research
• Carlo Zaniolo, Xin Zhou and Hyun J. Moon
• University of California, Los Angeles

2
Web Documents Version Management

• ltspec tstart 20000703 tend99991231gt
• ltw3c-doctype IsAttr yes tstart20000703
  tend20010627gt cr lt/w3c-doctypegt
• ltw3c-doctype IsAttr yes tstart20010627
  tend99991231gt rec lt/w3c-doctypegt
• ltstatus IsAttr yes tstart20000703
  tend20010627gt final lt/statusgt
• ltheader tstart20000703 tend99991231gt
• lttitle tstart20000703 tend99991231gt XML
  Linking Language (XLink)lt/titlegt
• ltversion tstart20000703 tend99991231gt
  Version 1.0lt/versiongt
• ltw3c-designation tstart20000703
  tend20010627gt WD-xlink-20000703
  lt/w3c-designationgt
• ltw3c-designation tstart20010627
  tend99991231gtREC-xlink-20010627
  lt/w3c-designationgt

ltspec w3c-doctypecr statusfinalgt
ltheadergt   lttitlegtXML Linking Language
(XLink)lt/titlegt   ltversiongtVersion
1.0lt/versiongt   ltw3c-designationgtWD-xlink-20000
703 lt/w3c-designationgt
Xlink Spec (Snapshot 1)
ltspec w3c-doctyperecgt ltheadergt   lttitlegtXML
Linking Language (XLink)lt/titlegt  
ltversiongtVersion 1.0lt/versiongt  
ltw3c-designationgtREC-xlink-20010627
lt/w3c-designationgt
Xlink Spec (Snapshot 2)
V-Document
3
ICAP (Incorporating Change Management into
Archival Process)

• Store Version history of an XML document as a new
  XML document, where
• for each element, vstart, vend attributes are
  added to timestamp its validity period
• temporally grouped model
• Complete history built from versions using XDIFF
• Support powerful historical queries by XQuery
• Example retrieve the history of Introduction
  section

for intro in doc ("v-xlinks.xml")/spec/body/div1
head"Introduction" return intro
4
Relational DB Temporal Information Management

• Timestamped tuples (projection without
  coalescing)

Temporally grouped history (projection with
coalescing)
5
XML Representation of RDB/XML Document History

• ltemployees tstart"1995-01-01" tend"1996-12-31"gt
• ltemployee tstart"1995-01-01" tend"1996-12-31"gt
• ltempno tstart"1995-01-01" tend"1996-12-31"gt100
  03lt/empnogt
• ltname tstart"1995-01-01" tend"1996-12-31"gtBoblt
  /namegt
• ltsalary tstart"1995-01-01" tend"1995-05-31"gt60
  000lt/salarygt
• ltsalary tstart"1995-06-01" tend"1996-12-31"gt70
  000lt/salarygt
• lttitle tstart"1995-01-01" tend"1995-09-30"gtEng
  ineerlt/titlegt
• lttitle tstart"1995-10-01" tend"1996-01-31"gtSr
  Engineerlt/titlegt
• lttitle tstart"1996-02-01" tend"1996-12-31"gtTec
  h Leaderlt/titlegt
• ltdept tstart"1995-01-01" tend"1995-09-30"gtQAlt/
  deptgt
• ltdept tstart"1995-10-01" tend"1996-12-31"gtRDlt/
  deptgt
• ltDOB tstart"1995-01-01" tend"1996-12-31"gt1945-
  04-09lt/DOBgt
• lt/employeegt
• lt!-- More --gt
• lt/employeesgt

6
ArchIS (Archival Information System) the
Architecture
A R C H I S
Temporal XML Data
H-views H-documents SQL/XML
Temporal XML Queries
H-tables
Active Rules/ update logs
7
ArchIS the Benefits

• Expressive temporal representations and data
  models using current standards
• Powerful languages for temporal queries using
  only current standards (and temporal functions)
• Indexing, clustering and query optimization
  techniques for efficient query support
• Significant performance benefits w.r.t native
  XML DBs, XML extensions in RDBMS, XTABLES ..

8
Mapping Temporal XQuery to SQL/XML

• Temporal projection retrieve the salary history
  of Bob

element salary_history for s in
doc("employees.xml")/
employees/employee/nameBob/salary return
s
select XMLElement (Name "salaryhistory",
XMLAgg (XMLElement (Name as "salary",
XMLAttributes (S.tstart as tstart,
S.tend as "tend"), S.salary))) from
employee_salary as S, employee_name as N where
N.id S.id and N.name 'Bob' group by N.id
9
Performance Study Query Performance
Q1,Q2 snapshot Q3,Q4 history Q5 temporal
slicing Q6 temporal join
10
Conclusion

• Temporal DBMS in standard XML/XQuery
• Transaction time, valid time and bitemporal
• Temporally grouped data model
• Uniform approach applicable to
• General Web/XML documents for content management
  and archiving
• Relational database temporal information
  management
• Best performance by mapping to SQL/XML