PB138 — XML databases, NoSQL
databases
NoSQL databases
• non relational databases, flexible schema
• often used for big data applications, clusters
• different storage structure than SQL databases
• give up constraints/transactions to improve performance
• low-level interface
NoSQL types
• key-value
◦ Redis, Memcached, Amazon SimpleDB…
• document (JSON, XML…)
◦ CouchDB, Elasticsearch, MongoDB…
• graph / RDF triple
◦ Virtuoso, Neo4j…
• object
◦ Caché, GemStone…
RDF databases / triple store
• standard data model (RDF)
• standardized interchange format (N-Triples, N-Quads, XML,…)
• query language (SPARQL), Linked Data
• native
◦ Apache Jena, Sesame/RDF4J…
• RDF layer to relational database
◦ Virtuoso, IBM DB2…
SPARQL
• SPARQL Protocol and RDF Query Language
• W3C Recommendation SPARQL 1.1, March 2013
1
• SELECT - values as table
• CONSTRUCT - extract RDF
• ASK - true/false
• DESCRIBE - extract RDF graph
• inferencing
SPARQL example
Ontology
ex1:FullProfessor rdf:subClassOf ex1:Professor.
ex1:AssistantProfessor rdf:subClassOf ex1:Professor.
ex1:Professor owl:equivalentClass ex2:Teacher
Data
ex1:Bob rdf:type ex1:FullProfessor .
ex1:Alice rdf:type ex1:AssistantProfessor .
ex2:Mary rdf:type ex2:Teacher
SPARQL example
Data
ex1:Bob rdf:type ex1:FullProfessor .
ex1:Alice rdf:type ex1:AssistantProfessor .
ex2:Mary rdf:type ex2:Teacher
SPARQL query
SELECT ?x
 WHERE {
 ?x rdf:type ex1:Professor
 }
• noone is Professor, but inferencing will find Bob, Alice, Mary
XML databases, when to use
• working with documents or metadata in XML format
• data format/schema changes over time
• complex and variable schema
2
• structure queries
XML database concepts
• basic element = document
• documents gathered in collections ("tables")
• query on document structure
• output is document, document fragment, or constructed XML
XML database types
• XML-enabled databases
◦ mapping XML data to own data model (relational, object…)
▪ character large object
▪ fragmented to series of tables/objects
▪ stored in XML Type
◦ ISO SQL/XML - element construction, data mapping, enhanced SQL with XQuery
◦ Oracle, IBM DB2, MS SQL, PostgreSQL
XML database types
• native XML databases
◦ using XML data model directly
• open-source
◦ eXist, Sedna, BaseX, MonetDB, Oracle/Berkeley DB XML
• commercial
◦ MarkLogic, Virtuoso, Qizx
Interface
XQuery
<titles>{
for $book in collection("books")/book where $book/year="1990"
return $book/title
}</titles>
3
Interface
XQuery Update
update delete collection("books")/book/isbn
update insert <title>XQuery Guide</title>
 into collection("books")/book[isbn="42"]
Interface
SQL/XML (result-table)
SELECT
  id, vol,
  xmlquery('$j/name', passing journal as "j") as name
FROM
  journals
WHERE
  xmlexists('$j[licence="CreativeCommons"]',
  passing journal as "j")
Interface
SQL/XML (result-XML)
SELECT XMLELEMENT (NAME "saleProducts",
  XMLNAMESPACES (DEFAULT 'http://posample.org'),
  XMLAGG (XMLELEMENT (NAME "prod",
  XMLATTRIBUTES (p.Pid AS "id"),
  XMLFOREST (p.name AS "name",
  i.quantity AS "numInStock"))))
FROM PRODUCT p, INVENTORY i
WHERE p.Pid = i.Pid
• XSLT - output transformation
• XML Schema - input validation
Interface
• XQJ - XQuery API for Java
◦ unified query layer between application and XML Datasource
4
◦ prepared statements
◦ binding variables
Interface
XQDataSource xqs = new ExistXQDataSource();
XQDataSource xqs = new SednaXQDataSource();
xqs.setProperty("serverName", "localhost");
XQConnection conn = xqs.getConnection();
XQExpression xqe = conn.createExpression();
String xqueryString = "for $x in doc('books.xml')//book
  return $x/title/text()";
XQResultSequence rs = xqe.executeQuery(xqueryString);
while(rs.next())
  System.out.println(rs.getItemAsString(null));
conn.close();
Interface
• XML:DB
• similar concept to JDBC, abstract interface to XML database
◦ Driver - access to given database
◦ Collection - document collection in database
◦ Services - support database features, e.g. XPathQueryService, XUpdateQueryService
◦ Resource - data stored in database
◦ ResourceSet - data as result of query
Storage
• intact document storage
◦ unique identifier for document
◦ preferably parse and index on storage
◦ on query: fast, if application need access to whole document and index can select the right
document
◦ slow, if additional parsing needed
Storage
• parsing documents
◦ document parsed on save and stored in own data model (eg. DOM)
5
◦ addressable numbered nodes - some operations faster based on numbers
◦ no need for parsing on query, more efficient
◦ retrieved document not 100% same
◦ fine granularity for addressing
◦ partial modifications
Indexing
• index scope - collection, database, document
• index target - document, node
• value index
◦ store all combinations of element/attribute values
• substring index
◦ for contains() etc., store n-grams
Indexing
• structural index
◦ track existing paths, enriched tree, trie, DataGuide etc.
Benchmark
• compare database performance
• mostly XQuery speed, less often Update
• data generator (up to GBs) and a set of XQueries
• XMark, XBench, XMach-1
• TPoX - complex database testing, XQuery and SQL/XML, indexing, XML Schema, XQuery Update
6