Share
Embed
Object Graph Mapping with Spring Data Neo4j 3
Slides accompanying a talk done by myself and Michael Hunger at Graph Connect 2013 http://graphconnect.com/gc2013 with video available at https://vimeo.com/80618455
Nicki Watt
November 20, 2013
Transcript
65 slides
-
Object Graph Mapping Spring Data Neo4j 3.0 Nicki
Wa( @ opencredo Michael Hunger @ neotechnology -
Agenda 1. Why Object-‐Graph-‐Mapping? 2. What is
Spring Data (Neo4j)? 3. Current State of affairs 4. SDN 3.0 5. The Future 6. Some Spring Data Neoj Use-‐Cases -
Why OGM ? -‐ Convenience -‐ Simplify
interac2on with underlying data source: Framework/Library handles basic CRUD type funcVonality for you -‐ Provides ability to work with na2ve domain language constructs: eg domain classes/objects (POJOs in Java) in your applicaVon -‐ Use Cases: examples at end of talk -
Why OGM ? But … -‐ IndirecVon, addiVonal
abstracVon layer adds performance, overhead -‐ Encourages users to pull larger parts of the database into memory to process instead of le_ng the db do its job -
Spring Data Neo4j -‐ OGM aimed at Java world
& uses Spring -‐ Neo4j Implementa2on of the Spring Data project: SpringSource/VMWare/Pivotal iniVaVve to give Spring developers easy access to the emerging world of NOSQL. Was the iniVal and first Spring Data project by Rod and Emil -‐ Simple programming model: annotaVon-‐based programming for applicaVons with rich domains -
Spring Data Neo4j -‐ features -‐ Two mapping modes:
Simple & Advanced -‐ Provides Spring Data Repository Support -‐ Ability to drop down to Neo4j Core API if required -‐ (Neo4j Server Support) -
SDN Domain Entities @NodeEntity @TypeAlias(value = "Conference")
public class Conference { @GraphId Long graphId; @Indexed String name; Date date; @RelatedTo(type="ATTENDEE") Set<Person> attendees; @RelatedTo(type="TALK") Set<Talk> talks; // . . . } -
SDN Domain Entities @NodeEntity @TypeAlias(value = "Person")
public class Person { @GraphId Long graphId; @Indexed(unique = true) String name; @RelatedToVia Iterable<Attended> talksAttended; // . . . } -
SDN Domain Entities (Relationship) @RelationshipEntity(type = "LISTENED_TO") public
class Attended { @GraphId Long graphId; @StartNode Person attendee; @EndNode Talk talk; Integer rating; // . . . } -
SDN Domain Entities @NodeEntity @TypeAlias(value = "Talk")
public class Talk { @GraphId Long graphId; @Indexed String name; String description; @RelatedTo(type="SPEAKER") Set<Speaker> speakers; @RelatedToVia(direction = INCOMING) Set<Attended> attendees; // . . . } -
SDN Domain Entities @TypeAlias(value = "Speaker") public class
Speaker extends Person { String bio; @RelatedTo(type="SPEAKER", direction = INCOMING) Iterable<Talk> talks; // . . . } -
SDN Repository Support import org....data.neo4j.repository.GraphRepository; import java.util.Set;
public interface TalkRepository extends GraphRepository<Talk> { Set<Talk> findTalksBySpeakersName(String name); } -
@NodeEntity @TypeAlias(value = "Talk") public class Talk {
@GraphId Long graphId; @Indexed String name; String description; @RelatedTo(type="SPEAKER") Set<Speaker> speakers; @RelatedToVia(direction = INCOMING) Set<Attended> attendees; // . . . } SDN Repository Support findTalksBySpeakersName -
SDN Repository Support findTalksBySpeakersName @TypeAlias(value = "Speaker")
public class Speaker extends Person { String bio; @RelatedTo(type="SPEAKER", direction = INCOMING) Iterable<Talk> talks; // . . . } @NodeEntity @TypeAlias(value = "Person") public class Person { @GraphId Long graphId; @Indexed(unique = true) String name; @RelatedToVia(type="ATTENDED") Iterable<Attended> talksAttended; // . . . } -
SDN Repository Support import org....data.neo4j.repository.GraphRepository; import java.util.Set;
public interface TalkRepository extends GraphRepository<Talk> { Set<Talk> findTalksBySpeakersName(String name); } START `talk_speakers`=node:`Person`(`name`={0}) MATCH (`talk`)-‐[:`SPEAKER`]-‐>(`talk_speakers`) RETURN `talk` -
SDN Repository Support import org....data.neo4j.repository.GraphRepository; import java.util.Set;
public interface TalkRepository extends GraphRepository<Talk> { // Repository Query Method @Query(value = "MATCH (talk:Talk)-‐[:SPEAKER]-‐>(speaker:Speaker) " + "WHERE speaker.name={0} RETURN talk") Set<Talk> findTalksBySpeakersNameQuery(String name); // . . . } -
Current State Of Affairs -‐ Current ProducVon Version: 2.3.2.RELEASE
-‐ Neo4j 1.9.3 support -
Disclaimer SDN 3.0 s)ll in progress ... •
Subject to change • Primary changes are Neo4j 2.0 compliance -‐ not finalised yet • follow progress on githubh"p:// spring.neo4j.org/source -
SDN 3.0 -‐ Base Neo4j 2.0 support -‐
Spring 3.2.5 (eventually 4 as well) support -‐ New Type Representa2on Strategy -‐ Labels -‐ Migra2on towards more Cypher driven founda2on -‐ Various bug fixes and enhancements -
Neo4j 2.0 Support -‐ Will be able to
run against a Neo4j 2.X DB -‐ Query using Labels -‐ Ability to use new Cypher Syntax -‐ Schema Indexes (Work in Progress) -
@NodeEntity @TypeAlias(value = "Person") public class Person {
@Query("MATCH (person)<-‐[:ATTENDEE]-‐(gathering:Conference)" + " WHERE id(person) = {self} " + " RETURN gathering") public Iterable<Conference> findAllConferencesAttended; // . . . } @NodeEntity @TypeAlias(value = "Conference") public class Conference { // . . . } -
@TypeAlias(value = "Speaker") public class Speaker extends Person {
String bio; @RelatedTo(type="SPEAKER", direction = INCOMING) Iterable<Talk> talks; // . . . } @NodeEntity @TypeAlias(value = "Person") public class Person { @GraphId Long graphId; @Indexed(unique = true) String name; @RelatedToVia(type="ATTENDED") Iterable<Attended> talksAttended; // . . . } -
Index vs Label TRS public interface ConferenceRepository extends
GraphRepository<Conference> { /* Query generated when using Index Based TRS START `conference`=node:__types__(className="Conference") WHERE `conference`.`date` = {0} RETURN `conference` Query generated when using Label Based TRS MATCH (`conference`:`Conference`) WHERE `conference`.`date` = {0} RETURN `conference` */ Set<Conference> findAllByDate(Date date); } -
-‐ Will allow for schema indexes** to be used
-‐ Will add addiVonal __TYPE__XX labels to nodes -‐ No addiVonal properVes need to be stored on nodes to represent hierarchy (If Index Based TRS used) -‐ No addiVonal nodes need to be created to represent hierarchy (If Sub Reference based TRS used) Label Type Representation Strategy -
Future Plans -‐ Cypher based • Use all Cypher
for the mapping layer • Align embedded and remote use • Benefit from Cypher enhancements • Leverage complex query mechanisms • Support more OGM type funcVonality • Client side caching • Lazy loading -
Future Plans -‐ OGM • PotenVally extract or use
standalone, cypher-‐based Java-‐Neo4j-‐OGM • Also useable without Spring (Data) -
Future Plans -‐ Query Results • Project query results
into object trees/graphs • Useful for direct view-‐layer integraVon • Less overhead by an intermediate domain model that has to be converted -
Future Plans -‐ Support • Be(er support for
• Spring Boot • Spring Data Rest • Remote transacVonal integraVon • JDBC-‐Driver in Spring -
Future Plans -‐ Migration • Migrate Type RepresentaVon
• Index-‐based -‐> Label based • Refactor/Rename EnVVes, Fields, References • Rename Indexes • @TypeAlias -
WhyOwnIt: Ownership Sharing Service • Sharing Culture
• You own things you only seldomly need • You need things you don‘t own • Match needs and available devices/tools • add raVngs, trust, recommendaVons • locaVon based, close to you for pick-‐up • Startup -
Music Sharing/Discovery/Recommendation Service • Recommend new arVsts and
songs to your users • take their previous listen-‐history into account • look at what‘s hot amongst their friends • infer broader selecVon via bands and genres • Detect duplicates using media-‐fingerprints • copyright infringements -
Other Use-‐Cases • Financial Services -‐ Asset / Account
Management • Bank -‐ Permission Management and AuthorizaVon • EnVtlement • Data Sopware Provider -‐ Impact analysis of changes • ImplicaVon of changes on the company core asset: DATA • Change tracking • Root Cause Analysis -
• Networked Storage Boxes • many components, disks, controllers,
power supply, sensors • Tracking of measurement data • Impact analyVcs • Early detecVon of anomalies • Intelligent fail over • Huge number of domain enVVes (~450) • generaVng domain classes and repositories SAN/NAS Provider -‐ Lifetime device tracking -
Game Retailer • Buy and Download & Game on
Demand • Fraud detecVon • delayed payments, charge backs, fradulent keys are invalid • Convert rules to traversals • co-‐occurrance of sudden large sales • fradulent e-‐mail pa(erns • CombinaVon with Oracle Infrastructure • Cross-‐Store -
Domain-‐Centric • Integrate in a Springframework Context •
ExisVng ApplicaVons with POJO Domains • Move a manageable amount of data in and out of the database into Domain Objects • Rich, connected domain • MulVple projecVons To SDN or Not to SDN ... -
Data-‐Centric • high volume database interacVons • reads
and writes (inserts) • thin domain view on top of graph queries • remote Client to Neo4j Server To SDN or Not to SDN ... -
• Spring Data Book • Spring Data Neo4j
Guidebook • Spring Data Neo4j on Spring.io • Chapter in Neo4j in AcVon • Cineasts.net Tutorial Resources -
Thanks ☺ Questions ? @techiewa(
@mesirii