Query Caching Techniques in XML Databases

1
Query Caching Techniques in XML Databases

• Kalpesh (02329027)

2
Outline

• Motivation
• Various Types of Caching
• How Semantic Caching Works ?
• Performance
• Conclusions
• References

3
Motivation

• Issues
• Behavior of Typical Query Engines
• Magnitude of Datasets Involved
• Diverse Geographical Locations of the Server
• Why we need Query Caching?
• Stores Query Results Locally.
• Reduces Execution Time.
• Reduces Burden on Server Side.

4
Types of Caching

• Page Caching
• Communication, in terms of Groups of Tuples.
• Tuple Caching
• Individual Tuple is fetched.
• Semantic Caching
• Dynamic Groups of Tuples are formed.
• Further Slides Talk about Semantic Caching.
• Implementation Improvement in Semantic Caching
  Based System XCache.

5
Benefits of Semantic Caching

• Reduces Space Overhead
• Regions (group of tuples) are dynamic
• Provides Complete flexibility in grouping tuples

6
Quilt An XML Query Language
ltResultgt FOR b IN dcument(books.xml")//bo
ok, a IN b/author WHERE a/firstname
"Crockett AND a/lastname "Johnson"
RETURN b lt/Resultgt

• The FOR clause generates bindings for one or more
  variables
• The LET clause binds a variable to the value of
  an expression
• The RETURN clause generates the output of the
  FLWR-expression

FLWR Expression
7
The Semantic Structure of Query
8
How it Works ?
9
Query comparison in Cache

• Disjoint Worst Case
• Totally Contained Best Case
• Query is broken for Remaining Two Cases

10
Example Query
11

• Union

• (Probe Query) U (Remainder Query) Original
  Query
• Probe Query Results are Fetched from Local
  Cache
• Remainder Query To be Shipped to Remote
  Database Server

12
Cache Region

• Contains Two Distinct Section
• Query Area
• Data Area
• Query Area Stores Variables, Conditions and
  Return Elements of the Query.
• All the Results from the past Queries are Stored
  into Data Area.

13
Performance

• Does this Approach really Improve Querying
  Performance ?
• Issues
• Overhead of Sending Query from Client to Server
• Execution of Query on the Remote Server
• Overhead of Fetching Results from Server to
  Client
• When Cache contains Partial Results,
• Ttotal Tship(query) Texecute(query)
  Tship(results)
• When Cache contains Total Results,
• Ttotal Texecute(query)

14
Implementation

• Limitation of Existing System
• Cache Size
• Query Replacement
• Work done
• Plugged in XCache System

15
Conclusions

• Talked about Various Types of Caching.
• Detailed look at Semantic Caching.
• Performance Issues.

16
References

• Semantic Cache Mechanism for Heterogeneous Web
  Querying. http//www8.org/w8-papers/3a-search-que
  ry/semantic/semantic.html
• Li Chen, Elke A. Rundensteiner, Song Wang.
  XCache - A Semantic Caching System for XML
  Queries.
• Home page for XCache System. http//davis.wpi.edu
  /dsrg/
• KWEELT - a framework to query XML data.
  http//cheops.cis.upenn.edu/Kweelt/N569