1. XML Stream Processing Yanlei Diao University of Massachusetts Amherst

2. 3/11/2016 Yanlei Diao, University of Massachusetts Amherst XML Data Streams  XML is the “wire format” for data exchanged online. Purchase orders http://www.oasis-open.org/committees/tc_home.php?wg_abbrev=ubl News feeds http://blogs.law.harvard.edu/tech/rss Stock tickers http://www.tickertech.com/products/xml/ Transactions of online auctions http://bigblog.com/online_auctions.xml Network monitoring data http://ganglia.sourceforge.net/

3. 3/11/2016 Yanlei Diao, University of Massachusetts Amherst XML Stream Processing  XML data continuously arrives from external sources, queries are evaluated every time a new data item is received.  A key feature of stream processing is the ability to process as it arrives. Natural fit for XML message brokering and web services where messages need to be filtered and transformed on-the-fly.  XML stream processing allows query execution to start before a message is completely received. Short delay in producing results. No need to buffer the entire message for query processing. Both are crucial when input messages are large, e.g. the equivalent of a database’s worth of data.

4. 3/11/2016 Yanlei Diao, University of Massachusetts Amherst Event-Based Parsing  XML stream processing is often performed on the granularity of an XML parsing event produced by an event-based API. Pub-Sub XML Processing Scalability < Start Document < Start Element: report < Start Element: section < Start Element: title Characters:Pub/Sub > End Element: title < Start Element: section < Start Element: figure < Start Element: title Characters: XML Processing > End Element: title > End Element: figure > End Element: section … > End Element: section > End Element: report > End Document

5. 3/11/2016 Yanlei Diao, University of Massachusetts Amherst Matching a Single Path Expression XFilter [ Altinel&Franklin’00 ], YFilter [ Diao et al.’02 ], Tukwila [ Ives et al.’02 ]  Simple paths: ( (“/” | “//”) (ElementName | “*”) )+ A simple path can be transformed to a regular expression Let  = set of element names: −‘/’ is translated to the ‘·’ concatenation operator −“//” is translated to  * −‘*’ is translated to  −“/a//b” can be translated to “a  * b”  A finite automaton (FA) for each path: mapping location steps to machine states.

6. 3/11/2016 Yanlei Diao, University of Massachusetts Amherst Query Compilation Location steps Automaton fragments /a //a /* a * a  * Map location steps to automaton fragments Concatenate automaton fragments for location steps in a query a * b  a * b  Query “/a//b” Is this Automaton deterministic or non-deterministic? For multi-path processing; o.w. not needed

7. 3/11/2016 Yanlei Diao, University of Massachusetts Amherst Event-Driven Query Execution  Query execution retrieves all matches of a path expression.  Event-driven execution of FA: Parsing events (esp. start of elements) drive the FA execution. Elements that trigger transitions to the accepting states are returned. figure  section * < Start Document < Start Element: report < Start Element: section < Start Element: title Characters:Pub/Sub > End Element: title < Start Element: section < Start Element: figure < Start Element: title Characters: XML Processing > End Element: title > End Element: figure > End Element: section … > End Element: section > End Element: report > End Document  Run FA over XML with nested structure, retrieve all results Approach 1: shred XML into a set of linear paths Approach 2: augment FA execution with backtracking

8. 3/11/2016 Yanlei Diao, University of Massachusetts Amherst Multi-Query Processing  Problem: evaluate a set Q = Q1, …, Qn of path queries against each incoming XML document.  Brute force: iterate the query set, one query at a time  Indexing of queries: inverse problem of traditional query processing Traditional DB: Data is stored persistently; queries executed against the data. Indexes of data enable selective searches of the data. XML stream processing: Queries are persistently stored; documents or their parsing events drive the matching of queries. Indexes of queries enable selective matching of documents to queries.  Sharing of processing: commonalities exist among queries. Shared processing avoids redundant work.

9. 3/11/2016 Yanlei Diao, University of Massachusetts Amherst YFilter [Diao et al.’03] builds a combined FA for all paths.  Complete prefix sharing among paths.  Moore machine with output: accepting states → partition of query ids.  Nondeterministic Finite Automaton (NFA)-based implementation: a small machine size, flexible, easy to maintain, etc. Constructing the Combined FSM a {Q1} b Q1=/a/b Q2=/a/c Q3=/a/b/c Q4=/a//b/c Q5=/a/*/b Q6=/a//c Q7=/a/*/*/c Q8=/a/b/c a {Q2} c c {Q3}  {Q4} c b * * c {Q5} c {Q6} * c {Q7} {Q3, Q8}

10. 3/11/2016 Yanlei Diao, University of Massachusetts Amherst YFilter uses a stack mechanism to handle XML.  Backtracking in the NFA.  No repeated work for the same element. An XML fragment … Execution Algorithm read 2 1 3 9 7 6 2 1 initial 1 Runtime Stack NFA match Q1 9 read 3 2 1 6 7 5 3 9 7 6 2 1 12 8 6 match Q3 Q8 10 11 Q5Q6Q4 c c b {Q1} {Q3, Q8} {Q2} {Q4} {Q6} {Q5} {Q7} a * c c * c c *  b 1 4 35 8 6 12 10 2 7 11 13 9

11. 3/11/2016 Yanlei Diao, University of Massachusetts Amherst Implementation Choices  Non-Deterministic Automata (NFA) versus Deterministic Automata (DFA) NFA: small machine, large numbers of transitions per element DFA: potentially large machine, one transition per element  Worse-case comparison for regular expressions: Single regular expression of length n m regular expressions compiled together Processing complexity Machine size Processing complexity Machine size NFAO(n2)O(n2)O(n)O(n)O(n2m)O(n2m)O(nm) DFAO(1)O(1) O(n)O(n) O(1)O(1) O(  nm ) YFilter specific O(3n) O(3nm) Possible in practice?  Restricted path expressions:

12. 3/11/2016 Yanlei Diao, University of Massachusetts Amherst Eager DFA  Green et al. studied the size of DFA for the restricted set of path expressions [Green et al.’03]  Eager DFA: Query compile time, translates from NFA to DFA Creates a state for every possible situation that runtime may require  Single path: Linear for “//e1/e2/e3/e4/e5” Exponential for “//e1/*/*/*/e5”  Multiple paths: Exponential for “//e1//b”, “//e2//b”, …, “//e5//b”

13. 3/11/2016 Yanlei Diao, University of Massachusetts Amherst Example of an Eager DFA The DFA size is O(2 w+1 ) where w is the number of ‘*’. //a/*/*/c/d “a a b b c d” “a b a b c d” Need to remember all the A’s in three consecutive characters, as different combinations of As may yield different results.

14. 3/11/2016 Yanlei Diao, University of Massachusetts Amherst Lazy DFA  Lazy DFA is constructed at run time, on demand Initially, it has a single state Whenever it attempts to make a transition into a missing state, it computes it and updates the transition  Hope: only a small set of the DFA states is needed.  Exploits DTD to derive upper bounds…

15. 3/11/2016 Yanlei Diao, University of Massachusetts Amherst Lazy DFA (contd.)  A DTD graph is simple, if the only loops are self-loops.  Theorem: the size of a lazy DFA is exponential only in the maximal number of simple cycles a path can intersect. Not exponential in # of paths! DTD graph  More complex recursive DTDs: e.g. “table” contains “list” and “list” contains “table” Even a lazy DFA grows large…

16. 3/11/2016 Yanlei Diao, University of Massachusetts Amherst Predicates in Path Expressions  Predicates can address attributes, text data, or positions of elements. Value of an attribute in an element, e.g., //section[@difficulty = “easy”]. Text data of an element, e.g., //section/title[text()=“XPath”]. Position of an element, e.g., //section/figure[text()=“XPath”][1].

17. 3/11/2016 Yanlei Diao, University of Massachusetts Amherst Predicate Evaluation  Extend the NFA including additional states representing successful evaluation of predicates and transitions to them  Potential problems A potentially huge increase in the machine size Destroy sharing of path expressions  Recent work [Gupta & Suciu 2003] : Possible to build an efficient pushdown automaton using lazy construction if No mixed content, such as “ 1 2 ”. Can afford to periodically rebuild the automaton from scratch. Can afford to train the automaton in each construction.

18. 3/11/2016 Yanlei Diao, University of Massachusetts Amherst YFilter: Mapping XML to Relational P1: //section//figure P2: //section/section/figure 452452725 path-tuple streams Shared Path Matching Engine  section * {P2} section figure {P1}  figure * 3 title section 4 figure 5 title 6 2 section 7 figure title 8 1 report  XQuery is much more complex than regular expressions.  Leverage efficient relational processing  XQuery stream processing = relational operations on pathtuple streams! parsing events

19. 3/11/2016 Yanlei Diao, University of Massachusetts Amherst Example Query Plan for FWR F: //section W1://section/title W2://section/figure/title R1://section//section//title R2://section//figure Q1: for $s in $doc//section[@difficulty=“easy”] where $s/title = “Pub/Sub” and $s/figure/title = “XML processing” return { $s//section//title } { $s//figure }   Π DupElim OuterJoin-Select Shared Path Matching Engine   FW1W2R1R2 An external (post-processing) plan for each query:  Selection: evaluates value-based predicates.  Projection: projects onto specific fields and removes duplicates.  Semijoin: handles correlations between for and where paths, finds query matches.  Outerjoin-Select: handles correlations between for and return paths, generates query results. Push all paths into the path engine. Π DupElim

20. 3/11/2016 Yanlei Diao, University of Massachusetts Amherst Buffering in XML Stream Processing  Buffering in XQuery stream processing Whether an element belongs to the result set is uncertain as it depends on predicates that have not been evaluated  Goal: avoid materialization and minimize buffering  Issues to address: What queries require buffering? What elements to buffer? When and how to prune dynamic data structures? −Buffered elements −State in any dynamic data structures

21. 3/11/2016 Yanlei Diao, University of Massachusetts Amherst Buffering in XML Stream Processing //section[title=“XML”]  Requires buffering? Yes  What element to buffer? section  When to prune? until the end of section is encountered, or when no more title can occur in this section. (Use DTD!) “XML processing” “Pub/Sub” “Scalability” @id=“intro” @difficulty=“easy” @source=“g1.jpg” @source=“g2.jpg” 3 title section 4 figure 5 title 6 2 section 7 figure title 8 1 report @difficulty=“easy”

22. 3/11/2016 Yanlei Diao, University of Massachusetts Amherst Buffering in XML Stream Processing //section[figure=“XML”]/title  Requires buffering? Yes  What element to buffer? title  When to prune? until a figure matches the predicate, or when no more figure can occur in this section (use DTD!), or the end of section is encountered. “XML processing” “Pub/Sub” “Scalability” @id=“intro” @difficulty=“easy” @source=“g1.jpg” @source=“g2.jpg” 3 title section 4 figure 5 title 6 2 section 7 figure title 8 1 report @difficulty=“easy”

23. 3/11/2016 Yanlei Diao, University of Massachusetts Amherst Buffering in XML Stream Processing  Requires buffering? Yes  What element to buffer? figure figure5?, figure7?  When to prune? until the predicate is true, or when no more title can occur in this section (use DTD!), or the end of section is encountered. “XML processing” “Pub/Sub” “Scalability” @id=“intro” @difficulty=“easy” @source=“g1.jpg” @source=“g2.jpg” 3 title section 4 figure 5 title 6 2 section 7 figure title 8 1 report @difficulty=“easy” //section[contains(.//title, “XML”)]//figure

24. 3/11/2016 Yanlei Diao, University of Massachusetts Amherst

25. 3/11/2016 Yanlei Diao, University of Massachusetts Amherst XQuery Usage Scenarios  XML message brokers Simple path expressions, for authentication, authorization, routing, etc. Single input message, relatively small Transient and streaming data (no indexes)  XML transformation language in Web Services Large and complex queries, mostly for transformation Input message + external data sources, small/medium sized data sets (xK -> xM) Transient and streaming data (no indexes)  Semantic data verification Mostly messages Potentially complex (but small) queries Streaming and multi-query optimization required

26. 3/11/2016 Yanlei Diao, University of Massachusetts Amherst Example 2 of an Eager DFA //a/b//c/d //e/f//g/h If there are l patterns with “//” per pattern, we need O(2^l) states to record the matching of the power set of the prefixes. The DFA size is O((x+1) l ), where x is the number of “//” per pattern, and l is the number of patterns.

27. 3/11/2016 Yanlei Diao, University of Massachusetts Amherst Full XQuery Stream Processor  Stream processing for the entire XQuery language [Daniela et al.’04] General algebra for XQuery Pull-based token-at-a-time execution model Lazy evaluation (like other functional languages) Some preliminary work on sharing [Diao et al.’04]  Buffering is a big concern [Barton et al.’03, Peng&Chawathe’03, Koch et al.’04]  Sharing is crucial for performance and scalability for large numbers of queries