Query Optimization Techniques and Performance Issues in XML and Parallel databases

1
Query Optimization Techniques and Performance
Issues in XML and Parallel databases

• CSE 8330
• Instructor Dr.Margaret H. Dunham
• Presenter Akshaya Aradhya

2
Topics to be covered

• Introduction
• Query optimization in XML databases
• Query optimization in Parallel databases
• Comparison
• Conclusion and Future work
• Bibliography

3
Introduction

• XML is an emerging standard for exchanging,
  storing and representing the data
• The data encoded in XML conforms to a DTD
  (Document Type Definition)
• XML structure is intuitive and it is easier to
  interpret it using its tree like structure.

4
Introduction

• XML data model is very complex when compared to
  other relational models, which renders a larger
  search space for optimizing XML queries
• In order to optimize XML queries, we need to
  study the equivalence issue related to the data
  and the query in order to find out the query
  equivalence before transforming the query

5
Introduction

• The techniques used to classify the XML query
  optimization techniques can be divided into
  groups based on the content and structure
• Content based query optimization Based on
  statistics or classification
• Query execution can be improved by classifying
  the elements, which transform the query based on
  constraints which are obtained from the data

6
Introduction

• The application of parallel database systems can
  be observed in decision support systems and a
  wide range of modern database applications.
• The machine architecture in parallel database
  systems are based on parallel dataflow
  architecture system, which make use of
  conventional, shared nothing hardware design.
• For each relation in the database, the tuples are
  de-clustered (partitioned) across disk storage
  units, which are attached to individual
  processors.

7
Introduction

• There are two properties demonstrated by
  parallelism, which makes it very desirable.
• The first one is called as linear scale-up, where
  the system can perform a task k times the size
  in a particular span of time, after the number of
  processors are increased by k.
• The second one is called as linear speedup where
  the response time is reduced by k times if we
  increase the number of processors by k times

8
Introduction

• During the query processing stage in parallel
  databases, parallelism can be exploited in three
  different ways.
• In the independent parallelism technique,
  different processors can execute different
  queries in parallel if the query operators do not
  depend on each other.
• By pipelining or by making use of inter-operator
  parallelism, the output of the producer to the
  consumer can be passed on in parallel by two or
  more operators in a producer consumer
  relationship.
• Finally, in intra-operator or partitioned
  parallelism technique, copies of the same query
  operator can be run on multiple processors
  simultaneously, where each of them can be
  operated on a partition of the data.

9
Optimization mechanism using ToXin tree

• ToXin indexing scheme was developed to overcome
  the limitation of applying optimization for path
  query processing.
• This scheme was developed with the primary goal
  of exploiting the path structure of the XML
  databases in all the stages of query processing.
• There are two types of index structures in Toxin
  called Value index and Path index

10
Optimization mechanism using ToXin tree

• Algorithm ConstructIndexTree
• Output Tree T
• ConstructIndexTree()
• 1. Perform a depth first traversal of the tree.
• 2. For each visited edge
• 2.1 Check whether the corresponding index
  edge has been added
• 2.1.1 For the current index edge of
  the XML element
• 2.1.1.1 Update the instance
  function in two redundant hash tables
  representing forward and backward
  navigation tables
• 2.1.1.2 Add the parent node
  and child node
• 2.2 If it has been added already, skip to
  the next index edge
• 3. Stop

11
Optimization mechanism in Lore

• An input query is divided into a set of sub
  queries where each operation is evaluated
  separately, as a part of the query.
• An effective execution order for these operations
  is obtained by creating evaluations for all the
  set of operations, which in turn helps in
  executing the queries faster.
• The final result can be obtained by joining all
  the aggregation of the results together.

12
Optimization mechanism in Lore

• Algorithm PlanSelectionAlgorithm
• Input Input list (for the query)
• Output Plan P
• PlanSelectionAlgorithm (input list)
• 1. Create a structure in order to track the
  binding variables
• 2. while input list is not empty
• 2.1 For each element in the input list
• 2.1.1 Based on the current bound
  variables, find the cheapest access method for
  the remaining steps
• 2.1.2 If the step has the least cost,
  mark the variables as bound and add it to the
  plan P
• 2.1.3 Remove the chosen step
• 3. Return the final plan P obtained from the
  previous steps

13
Optimizing queries in XML structured document
databases

• Using a set oriented algebraic technique named
  PAT algebra, a series of set related operations
  and rules are defined.
• PAT expressions are obtained by transforming
  input queries, after checking for the correctness
  of their syntax.
• Based on the relationship of elements in the DTD,
  the PAT expressions can be normalized with the
  help of the PAT algebra in order to get a new
  query.

14
Query optimization based on Schema
15
Query optimization by pruning and rewriting
queries
16
Query optimization by classification of elements
17
Join Strategy Selection
18
Optimal Serial Plan (in identical processors)
19
Comparison between Relational Database Management
System vs. XML Database System
20
(No Transcript)
21
Comparison of algorithms
22
Conclusion and Future work

• The tree generation algorithm and some of the
  optimal plan selection and generation algorithms
  run in polynomial time and hence, they need to be
  optimized to run in linear time.
• PAT algebra is being extended to make it more
  suitable for query optimization. Frequency search
  operations heavily make use of the indexing
  techniques in PAT.
• The future research will also be focused more
  towards generation and use of partially
  correlated sub-plans, which depend on bindings
  passed between portions of query plan.
• When a significant number of paths pass through a
  small number of objects, a transformation which
  introduces a group-by clause can be useful.
• Further examination is being conducted in order
  to implement the Toxin Graph and to check if the
  Toxin Tree can be extended to be used as an
  alternative to DOM for querying, updating and
  storing XML documents

23
Conclusion and Future work

• Value based grouping and join techniques are
  being investigated along with multi-way
  structural joins, new access methods for merged
  operators and several structural pattern
  techniques.
• In addition to this, new optimization algorithms
  have to be implemented to improve caching in Web
  Service Management Systems, XQuery language
  constructs are to be optimized.
• Cost based decisions are to be integrated in
  earlier stages of the query evaluation process
  and the cost model has to be refined in order
  model the CPU cost in a precise manner.

24
Bibliography

• 1 Dunren Che, Karl Aberer, and Tamer. 2006.
  Query optimization in XML structured-document
  databases. The VLDB Journal 15, 3 (September
  2006), 263-289.
• 2 Jason McHugh and Jennifer Widom. 1999. Query
  Optimization for XML. In Proceedings of the 25th
  International Conference on Very Large Data Bases
  (VLDB '99), Malcolm P. Atkinson, Maria E.
  Orlowska, Patrick Valduriez, Stanley B. Zdonik,
  and Michael L. Brodie (Eds.). Morgan Kaufmann
  Publishers Inc., San Francisco, CA, USA, 315-326.
• 3 Boag, S. Berglund, A. Chamberlin, D.
  Siméon, J. Kay, M. Robie, J. Fernández, M. F.
  (2007), 'XML Path Language (XPath) 2.0' ,
  Technical report, W3C , http//www.w3.org/TR/2007/
  REC-xpath20-20070123/ .
• 4 Haw, S.C and Rao, G.S.V.R.K., 2005. Query
  Optimization Techniques for XML Databases.
  International Journal of Information Technology,
  2(1) 97 104.
• 5 S. Groppe and S. Bottcher Schema-based
  Query Optimization for XQuery Queries,
  Proceedings of the Advances in Databases and
  Information Systems 2005, Tallinn, Estonia, 2005.
• 6 Mary F. Fernandez and Dan Suciu. 1998.
  Optimizing Regular Path Expressions Using Graph
  Schemas. In Proceedings of the Fourteenth
  International Conference on Data Engineering
  (ICDE '98). IEEE Computer Society, Washington,
  DC, USA, 14-23.
• 7 Dung Xuan Thi Le, Stephane Bressan, David
  Taniar, and Wenny Rahayu. 2007. Semantic XPath
  query transformation opportunities and
  performance. In Proceedings of the 12th
  international conference on Database systems for
  advanced applications (DASFAA'07), Ramamohanarao
  Kotagiri, P. Radha Krishna, Mukesh Mohania, and
  Ekawit Nantajeewarawat (Eds.). Springer-Verlag,
  Berlin, Heidelberg, 994-1000.

25
Bibliography

• 8 Atri Salminen and Frank Wm. TompaPat
  expressions an algebra for text search. In Acta
  Linguista Hungarica 41, pages 277 306, 1994.
• 9 F.Rizzolo and A.Mendelzon. Indexing XML Data
  with ToXin. In Proc. 4th Int. Workshop on the Web
  and Database (in Conjunction with ACM SIGMOD),
  Santa Barbara, CA, May 2001.
• 10 Jason McHugh and Jennifer Widom Query
  Optimization for XML. In proceedings of the 25th
  Very Large Data Bases Conference, Edinburgh,
  Scotland, 1999.
• 11 Wei Sun Daxin Liu Wansong Zhang , "An
  efficient method for XML queries optimization
  based DTD abstraction and classification,"
  Intelligent Control and Automation, 2004. WCICA
  2004. Fifth World Congress on , vol.5, no., pp.
  3926- 3929 Vol.5, 15-19 June 2004
• 12 Alberto O. Mendelzon. ToX The Toronto XML
  Server. Proc. Int. Database Engineering and
  Applications Symposium (IDEAS). IEEE CS Press.
  Edmonton, Canada, July 2002.
• 13 J. McHugh, S. Abiteboul, R. Goldman, D.
  Quass, and J. Widom. Lore A Database Management
  System for Semistructured Data. SIGMOD Record,
  26(3)54-66, September 1997.
• 14 McHugh, J., Widom. J., 1999b. Optimizing
  branching path expressions. Technical Report,
  Stanford University.
• 15 Ke Geng, Gillian Dobbie, and Yulong Meng.
  2009. Survey of XML Semantic Query Optimization.
  In Proceedings of the 2009 Fourth International
  Conference on Internet Computing for Science and
  Engineering (ICICSE '09). IEEE Computer Society,
  Washington, DC, USA, 297-300.
• 16 Tae-Sun Chung and Hyoung-Joo Kim. 2002.
  Extracting indexing information from XML DTDs.
  Inf. Process. Lett. 81, 2 (January 2002), 97-103.
  
• 17 Wu, Y., Patel, J.M., Jagadish, H.V.
  Structural join order selection for XML query
  optimization. In ICDE, pp. 443-454. IEEE
  Computer Society, New York (2003)

26
Bibliography

• 18 Abdelkader Hameurlain, Franck Morvan
  Evolution of Query Optimization Methods. T.
  Large-Scale Data- and Knowledge-Centered Systems
  1 211-242 (2009)
• 19 Andreas M. Weiner, Theo Härder An
  integrative approach to query optimization in
  native XML database management systems. IDEAS
  2010 64-74
• 20 Amol Deshpande and Lisa Hellerstein. 2008.
  Flow Algorithms for Parallel Query Optimization.
  In Proceedings of the 2008 IEEE 24th
  International Conference on Data Engineering
  (ICDE '08). IEEE Computer Society, Washington,
  DC, USA, 754-763.
• 21 S. M. Mahajan and V. P. Jadhav. 2011. A
  survey of issues of query optimization in
  parallel databases. In Proceedings of the
  International Conference Workshop on Emerging
  Trends in Technology (ICWET '11). ACM, New York,
  NY, USA, 553-554.
• 22 Sai Wu, Feng Li, Sharad Mehrotra, and Beng
  Chin Ooi. 2011. Query optimization for massively
  parallel data processing. In Proceedings of the
  2nd ACM Symposium on Cloud Computing (SOCC '11).
  ACM, New York, NY, USA, , Article 12 , 13 pages.
• 23 David J. DeWitt and Jim Gray. 1990. Parallel
  database systems the future of database
  processing or a passing fad?. SIGMOD Rec. 19, 4,
  104-112.
• 24 Ashish Thusoo, Joydeep Sen Sarma, Namit
  Jain, Zheng Shao, Prasad Chakka, Suresh Anthony,
  Hao Liu, Pete Wyckoff, and Raghotham Murthy.
  2009. Hive a warehousing solution over a
  map-reduce framework. Proc. VLDB Endow. 2, 2
  (August 2009), 1626-1629.
• 25 Foto N. Afrati and Jeffrey D. Ullman. 2010.
  Optimizing joins in a map-reduce environment. In
  Proceedings of the 13th International Conference
  on Extending Database Technology (EDBT '10),
  Ioana Manolescu, Stefano Spaccapietra, Jens
  Teubner, Masaru Kitsuregawa, Alain Leger, Felix
  Naumann, Anastasia Ailamaki, and Fatma Ozcan
  (Eds.). ACM, New York, NY, USA, 99-110.
• 26 Utkarsh Srivastava, Kamesh Munagala,
  Jennifer Widom, and Rajeev Motwani. 2006. Query
  optimization over web services. In Proceedings of
  the 32nd international conference on Very large
  data bases (VLDB '06), Umeshwar Dayal, Khu-Yong
  Whang, David Lomet, Gustavo Alonso, Guy Lohman,
  Martin Kersten, Sang K. Cha, and Young-Kuk Kim
  (Eds.). VLDB Endowment 355-366.