PGQL: A Language for Graphs

Copyright © 2018, Oracle and/or its affiliates. All rights reserved.
Oracle Ask TOM Office Hours:
A Language for Graphs: PGQL
Albert Godfrind, Solutions Architect @agodfrin
Oskar Van Rest @oskarvanrest
Jean Ihm, Product Manager @JeanIhm

Copyright © 2017, Oracle and/or its affiliates. All rights reserved. |
AskTOM sessions on property graphs
• Today’s is the fifth session on property graphs
– February’s session introduced Property Graphs
– March's session explained how to model graphs from relational data
– May's session explained how to perform graph analytics
– August's session showed how we could view graphs
– In case you missed them, recordings are available at the URL above
• Today’s topic: PGQL: A language for Graphs
• Visit the Spatial and Graph landing page to view recordings; submit
feedback, questions, topic requests; view upcoming session dates and
topics; sign up
2
https://devgym.oracle.com/pls/apex/dg/office_hours/3084

3
The Story So Far …

Oracle Big Data Spatial and Graph
• Available for Big Data platform/BDCS
– Hadoop, HBase, Oracle NoSQL
• Supported both on BDA and commodity
hardware
– CDH and Hortonworks
• Database connectivity through Big Data
Connectors or Big Data SQL
• Included in Big Data Cloud Service
Oracle Spatial and Graph
• Available with Oracle 18c/12.2/DBCS
• Using tables for graph persistence
• Graph views on relational data
• In-database graph analytics
– Sparsification, shortest path, page rank, triangle
counting, WCC, sub graphs
• SQL queries possible
• Included in Database Cloud Service
4
Graph Product Options

Graph Storage Management
5
Architecture of Property Graph
Graph Analytics
Blueprints/Tinkerpop/Gremlin
REST/WebService
APIs
Java,Groovy,Python,…
Scalable and Persistent Storage
Oracle NoSQL
Database
Oracle RDBMS Apache HBase
What we focus
on now
Parallel In-Memory Graph Analytics
pgql> SELECT ...

• What is a graph?
– Data model representing entities as
vertices and relationships as edges
– Optionally including attributes
• Flexible data model
– No predefined schema, easily extensible
– Particularly useful for sparse data
• Enabling new kinds of analytics
– Overcoming limitations in relational
technology
6
Graph Data Model
E
A D
C B
F

Graph Analysis for Business Insight
7
Identify
Influencers
Discover Graph Patterns
in Big Data
Generate
Recommendations

8
Executing PGQL

Invoking Graph Analytics
1. Get an in-memory analyst
2. Read the graph from the data store into memory
3. Run analytics and execute PGQL statements
9
instance = Pgx.getInstance();
session = instance.createSession();
analyst = session.createAnalyst();
pg = session.readGraphWithProperties(...)
analyst.pageRank(pg,...)
pg.queryPgql ("SELECT ... MATCH ... WHERE ...")

Running PGQL queries
• Execute a PGQL query:
• Print the results:
• Iterate through the results:
• Close the result set:
11
rs = pg.queryPgql("SELECT ... MATCH ... WHERE ...")
rs.print() // prints all results
rs.print(10) // prints the first 10 results
for (PgqlResult result : rs) {
String movie_title = result.getString("m.title");
String movie_budge = result.getFloat("m.budget");
...
}
rs.close()

Using Bind Variables
• Much like JDBC: use "?" for variable markers.
• Prepare the Statement:
• Set bind values
• Execute prepared statement
12
stmt = pg.preparePgql("SELECT ... MATCH ... WHERE ...")
stmt.setString(1, "Alibaba") // first bind variable has index 1
stmt.setInt(2, 100)
stmt.setDouble(3, 2898.10)
...
rs = stmt.executeQuery()

• Open Source
• Originally for biological research
• Now a general platform for
complex graph analysis and
visualization
• Desktop / Pure Java
• Extensible via Plug-ins
• Also Javascript library
(cytoscape.js)
13
http://www.cytoscape.org

Type the query or
load from a file
Select the properties
needed

15
Deep Dive into PGQL

PGQL – Property Graph Query Language
16
SELECT p2.name AS friend_of_friend
FROM facebook_graph
MATCH (p1:Person) -/:friend_of{2}/->
(p2:Person)
GROUP BY ..
ORDER BY ..
LIMIT ..
OFFSET ..
Example query:• Core Features
– SQL-like syntax
• SELECT .. FROM .. WHERE ..
• Grouping and aggregation: GROUP BY, COUNT, AVG,
MIN, MAX, SUM
• Sorting: ORDER BY, ASC, DESC
– Graph pattern matching
• Define a high-level pattern and match all the
instances in the data graph
– Recursive path expressions
• Often recursive in nature
– E.g. can I reach from vertex A to vertex B via some
number of edges?
• Use cases: detecting circular cash flow (fraud
detection), network impact analysis, etc.

Example: Network Impact Analysis
• How does network disruption impacts
reachability between electric devices?
17
Device
OPEN
OPEN
OPEN
OPEN
OPEN
OPENOPEN
OPEN
OPEN OPEN
OPEN
OPEN OPEN
OPEN OPENCLOSED
CLOSED
source deviceElectric_Network
OPEN
OPEN
Connector
Connection
PATH connects_to
AS (from) <-[c1]- (connector) -[c2]-> (to)
WHERE c1.status = 'OPEN'
AND c2.status = 'OPEN'
SELECT n.nickname, COUNT(m)
FROM Electric_Network
MATCH (n:Device) -/:connects_to*/-> (m:Device)
WHERE java_regexp_like(n.nickname, 'Regulator')
AND n <> m
GROUP BY n
ORDER BY COUNT(m) DESC, n.nickname
+--------------------------+----------+
| n.nickname | COUNT(m) |
+--------------------------+----------+
| Regulator, VREG2_A | 1596 |
| Regulator, VREG4_B | 1537 |
| Regulator, VREG4_C | 1537 |
| Regulator, HVMV_Sub_RegA | 3 |
| Regulator, HVMV_Sub_RegB | 3 |
+--------------------------+----------+
Query: For each ‘Regulator’ device,
show number of reachable devices
following only ‘OPEN’ connections.
Example result

PGQL 1.0
Regular Path Queries
Comparison to SQL
18
WITH temp(device_id, device_name) AS (
-- Anchor member:
SELECT device_id, name
FROM Devices
WHERE name = 'Regulator, HVMV_Sub_RegB‘
UNION ALL
-- Recursive member:
SELECT Devices.device_id, Devices.name
FROM temp, Devices, Connections conn1,
Connections conn2, Connectors
WHERE temp.device_id = conn1.to_device_id
AND conn1.from_connector_id = Connectors.connector_id
AND Connectors.connector_id = conn2.from_connector_id
AND conn2.to_device_id = Devices.device_id
AND temp.device_id != Devices.device_id)
CYCLE device_id SET cycle TO 1 DEFAULT 0
SELECT DISTINCT device_name
FROM temp
WHERE cycle = 0
AND device_name <> 'Regulator, HVMV_Sub_RegB'
PATH connects_to := (from) <- (connector) -> (to)
SELECT y.name
MATCH (x:Device) -/:connects_to*/-> (y:Device),
WHERE x.name = 'Regulator, HVMV_Sub_RegB'
AND x <> yPGQL
SQL
Query:
Which devices are connected
transitively to device
'Regulator, HVMV_Sub_RegB‘?

Basic graph pattern matching
SELECT v3.name, v3.age
FROM socialNetworkGraph
MATCH (v1:Person) –[:friendOf]-> (v2:Person) –[:knows]-> (v3:Person)
WHERE v1.name = ‘Amber’
Query: Find all people who are known
by friends of ‘Amber’.
socialNetwork
Graph
100
:Person
name = ‘Amber’
age = 25
200:Person
name = ‘Paul’
age = 30
300
:Person
name = ‘Heather’
age = 27
777
:Company
name = ‘Oracle’
location =
‘Redwood City’
:worksAt{1831}
startDate = ’09/01/2015’
:friendOf{1173}
:knows{2200}
:friendOf {2513}
since = ’08/01/2014’
19
• Find all instances of a given pattern/template in the data graph

Basic graph pattern matching
SELECT v3.name, v3.age
FROM socialNetworkGraph
MATCH (v1:Person) –[:friendOf]-> (v2:Person) –[:knows]-> (v3:Person)
WHERE v1.name = ‘Amber’
Query: Find all people who are known
by friends of ‘Amber’.
socialNetwork
Graph
100
:Person
name = ‘Amber’
age = 25
200:Person
name = ‘Paul’
age = 30
300
:Person
name = ‘Heather’
age = 27
777
:Company
name = ‘Oracle’
location =
‘Redwood City’
:worksAt{1831}
startDate = ’09/01/2015’
:friendOf{1173}
:knows{2200}
:friendOf {2513}
since = ’08/01/2014’
20
• Find all instances of a given pattern/template in the data graph

Regular path expressions
21
• Matching a pattern
repeatedly
– Define a PATH expression at the
top of a query
– Instantiate the expression in the
MATCH clause
– Match repeatedly, e.g. zero or
more times (*) or one or more
times (+)
:Person
name = ‘Amber’
age = 29
100
:Person
name = ‘Dwight’
age = 15
400
:Person
name = ‘Paul’
age = 64
200
:Person
name = ‘Retta’
age = 43
300
:Person
name = ‘Peter’
age = 12
500
0 1
:likes
since = ‘2016-04-04’
:likes
since = ‘2016-04-04’
2
4
3
5
7
6
:likes
since = ‘2013-02-14’
:likes
since = ‘2015-11-08’
:has_father
:has_father
:has_mother
:has_mother
snGraph
PATH has_parent AS (child) –[:has_father|has_mother]-> (parent)
SELECT x.name, y.name, ancestor.name
FROM snGraph
MATCH (x:Person) –/:has_parent+/-> (ancestor)
, (y) -/:has_parent+/-> (ancestor)
WHERE x.name = 'Peter' AND x <> y

Regular path expressions
22
• Matching a pattern
repeatedly
– Define a PATH expression at the
top of a query
– Instantiate the expression in the
MATCH clause
– Match repeatedly, e.g. zero or
more times (*) or one or more
times (+)
:Person
name = ‘Amber’
age = 29
100
:Person
name = ‘Dwight’
age = 15
400
:Person
name = ‘Paul’
age = 64
200
:Person
name = ‘Retta’
age = 43
300
:Person
name = ‘Peter’
age = 12
500
0 1
:likes
since = ‘2016-04-04’
:likes
since = ‘2016-04-04’
2
4
3
5
7
6
:likes
since = ‘2013-02-14’
:likes
since = ‘2015-11-08’
:has_father
:has_father
:has_mother
:has_mother
snGraph
Result set
PATH has_parent AS (child) –[:has_father|has_mother]-> (parent)
SELECT x.name, y.name, ancestor.name
FROM snGraph
MATCH (x:Person) –/:has_parent+/-> (ancestor)
, (y) -/:has_parent+/-> (ancestor)
WHERE x.name = 'Peter' AND x <> y
+--------+--------+---------------+
| x.name | y.name | ancestor.name |
+--------+--------+---------------+
| Peter | Retta | Paul |
| Peter | Dwight | Paul |
| Peter | Dwight | Retta |
+--------+--------+---------------+

SELECT, FROM, MATCH, WHERE
• SELECT creates tabular projection of results
– Alias names: SELECT (x.prop1 + x.prop2) / 2 AS avg_prop1_prop2
– Remove duplicates: SELECT DISTINCT x.prop
• FROM and MATCH to specify the graph and the pattern to match on it
– Example: FROM g MATCH (n) -[e]-> (m)
• WHERE (optional) contains a filter expression:
– The filter expression typically involves properties of vertices/edges defined in the MATCH clause
• e.g. WHERE m.budget > 1000000 AND m.budget < 2000000 AND …
23
Operator type Operator Example PGQL
Arithmetic +, -, *, /, % 1 + 1
Relational =, <>, <, >, <=, >= 3 < 4
Logical AND, OR, NOT true AND NOT false

Patterns, labels, regular expressions
24
Syntax Description
-/:lbl/-> Matches if there exists a path that connects the source and
destination by one match of the given pattern
-/:lbl*/-> Matches if there exists a path that connects the source and
destination by zero or more matches of the given pattern
-/:lbl+/-> Matches if there exists a path that connects the source and
destination by one or more matches of the given pattern
-/:lbl?/-> Matches if there exists a path that connects the source and
destination by zero or one matches of the given pattern
-/:lbl{2}/-> Matches if there exists a path that connects the source and
destination by exactly 2 matches of the given pattern
-/:lbl{2,}/-> Matches if there exists a path that connects the source and
destination by at least 2 matches of the given pattern
-/:lbl{2,3}/-> Matches if there exists a path that connects the source and
destination by at least 2 and at most 3 (inclusive) matches of
the given pattern
-/:lbl{,3}/-> Matches if there exists a path that connects the source and
destination by at least 0 and at most 3 (inclusive) matches of
the given pattern
Syntax Description
(v1) Match vertex v1
-[e1]-> Match outgoing edge e1
<-[e1]- Match incoming edge e1
-[e1]- Match incoming or outgoing edge e1
() Match anonymous vertex
->
<-
-
Match anonymous
outgoing/incoming/any-directional
edge
(v1:lbl1|lbl2) Match vertex v1 with label lbl1 or lbl2
-[:lbl1]-> Match anonymous outgoing edge with
label lbl1

Grouping and aggregation
• GROUP BY creates groups of intermediate solutions
• COUNT, MIN, MAX, SUM, AVG to compute aggregations over groups of solutions
– Use DISTINCT to eliminate duplicates before aggregation
• E.g. COUNT(DISTINCT n.name)
– Omit GROUP BY to aggregate over the entire set of solutions (returns a single row)
• HAVING filters entire groups of solutions
• E.g. HAVING COUNT(*) > 10
25
SELECT m.title, MAX(e.time)
FROM "Movies Graph"
MATCH (m) <-[e:click]- ()
WHERE m.type = 'movie'
GROUP BY m
Query: “for each movie, return the
time of the most recent click”
+---------------+---------------------+
| m.title | MAX(e.time) |
+ --------------+---------------------+
| "Yes Man" | 2011-08-25T20:19:40 |
| "Match Point" | 2010-12-17T02:42:32 |
+---------------+---------------------+

ORDER BY, LIMIT, OFFSET
• ORDER BY
– Sort result ascendingly (default): ORDER BY n.prop ASC, …
– Sort result descendingly: ORDER BY n.prop DESC, …
• LIMIT and OFFSET
– For returning the “top-k” results or for “pagination” of results
• E.g. OFFSET 200 LIMIT 100 returns results 200 to 300
26
200100
500
300
400
type : ‘movie’
title : ‘Match Point’
type : ‘customer’
first_name : ‘Iksana’
type : ‘movie’
title : ‘Yes Man’
first_name : ‘Maria’
first_name : ‘Franklin’
click 0
click 1
click 2
click 3
time : 2011-01-01T20:57:21
time : 2010-12-17T02:42:32
time : 2011-08-25T20:19:40
time : 2012-09-29T02:49:09
Movies Graph
SELECT n.first_name
FROM "Movies Graph"
MATCH (n)
WHERE n.type = 'customer'
ORDER BY n.first_name
LIMIT 2
+--------------+
| n.first_name |
+--------------+
| Franklin |
| Iksana |
+--------------+
Query: “order customers
by first name and return
first two results”

PGQL data types in the In-Memory Analyst (PGX)
(note: for PGQL-to-SQL, the available data types correspond to those in the RDBMS)
27
Data type Name in PGX config Example PGQL literal Implicitly converts to
(e.g. integer + long => LONG)
STRING string 'Franklin'
INTEGER integer - LONG, FLOAT, DOUBLE
LONG long 2000 FLOAT, DOUBLE
FLOAT float - DOUBLE
DOUBLE double 1.95
BOOLEAN boolean true
DATE local_date
(note: don’t use deprecated date)
DATE '2017-08-18'
TIME time TIME '20:15:00' TIME WITH TIME ZONE
TIMESTAMP timestamp TIMESTAMP '2017-08-18 20:15:00' TIMESTAMP WITH TIME ZONE
TIME WITH TIME ZONE time_with_timezone TIME '20:15:00+08' TIME
TIMESTAMP WITH TIME ZONE timestamp_with_timezone TIMESTAMP '2017-08-18 20:15:00+08' TIMESTAMP

Explicit type conversion (CAST)
28
• Convert from string:
– CAST('true' AS BOOLEAN) => true
– CAST('2.50' AS DOUBLE) => 2.50
– CAST('2017-08-18' AS DATE) => DATE '2017-08-18‘
• Convert to string:
– CAST(true AS STRING) => 'true'
– CAST(2.50 AS STRING) => '2.50'
– CAST(DATE '2017-08-18' AS STRING) => '2017-08-18'
• Narrowing numeric conversion
– CAST(1.2 AS INTEGER) => 1
• Extract date or time portion from a
timestamp:
– CAST(TIMESTAMP '2017-08-18 20:15:00' AS DATE)
=> DATE '2017-08-18'
– CAST(TIMESTAMP '2017-08-18 20:15:00' AS TIME)
=> TIME '20:15:00'
• Create a timestamp from a date or a
time:
– CAST(DATE '2017-08-18' AS TIMESTAMP)
=> TIMESTAMP '2017-08-18 00:00:00'
– CAST(TIME '20:15:00' AS TIMESTAMP)
=> TIMESTAMP '2018-03-19 20:15:00'
It takes the current date
to create a timestamp

Built-in functions
29
Function Description Return type
id(vertex/edge) get the identifier of a vertex or edge exact numeric /
string
has_label(vertex/edge, lbl) returns true if the vertex/edge has the
label; false otherwise
boolean
labels(vertex) returns the labels of a vertex set<string>
label(edge) returns the label of an edge string
all_different(value1,
value2, value3, …)
returns true if the values are all different,
e.g. all_different(1, 5, 2)  true
e.g. all_different(1, 5, 1)  false
boolean
in_degree(vertex) get outgoing number of edges exact numeric
out_degree(vertex) get incoming number of edges exact numeric
java_regexp_like(string,
pattern)
returns whether the string matches the
pattern1
boolean
1 see https://docs.oracle.com/javase/7/docs/api/java/util/regex/Pattern.html
•
Use all_different to avoid
explosion of non-equality constraints:
SELECT COUNT(*)
MATCH (p1:Person) -[:likes]-> (friend:Person)
, (p2:Person) -[:likes]-> (friend)
WHERE p1 <> p2 AND p1 <> p3 AND p2 <> p3
SELECT COUNT(*)
MATCH (p1:Person) -[:likes]-> (friend:Person)
WHERE all_different(p1, p2, p3)

• EXISTS tests for the existence of a pattern, given the already obtained bindings
EXISTS subqueries
30
SELECT DISTINCT friendOfFriend.name
FROM g
MATCH (a:Person) –[:friend_of]-> () –[:friend_of]-> (friendOfFriend)
WHERE a.name = 'Peter'
AND NOT EXISTS (
SELECT *
MATCH (a) –[:friend_of]-> (friendOfFriend)
)
SELECT friendOfFriend.name
MATCH (a:Person) –/:friend_of{2}/-> (friendOfFriend)
WHERE a.name = 'Peter'
Query: “Find friends of
friends of Peter”
More concise way of
expressing the query
•
Instead of using SELECT DISTINCT it’s sometimes faster to test for path existence using -/../-> :

EXISTS subquery in PATH expression
31
PATH calls_but_no_global_write AS (:Function) –[:calls]-> (dst:Function)
WHERE NOT EXISTS (
SELECT *
MATCH (dst) -/:calls*/-> (:Function) –[:writes]-> (:GlobalVariable)
)
SELECT f2.name
FROM codeBaseGraph
MATCH (f1:Function) -/:calls_but_no_global_write+/-> (f2)
WHERE f1.name = 'func1'
Query: “In a code base, find functions that are reachable from 'func1' such
that no function along the “calls path” transitively writes to a global variable”

Scalar subqueries
32
Example query (from the
LDBC graph benchmark)
SELECT person.id
, COUNT(*) AS messageCount
, SUM((SELECT COUNT(r)
MATCH (message)<-[r:replyOf]-(comment:comment))
) AS replyCount
, SUM((SELECT COUNT(l)
MATCH (:person)-[l:likes]->(message))
) AS likeCount FROM ldbcSnbGraph
MATCH (tag:tag) <-[:hasTag]- (message:post|comment)
, (message) -[:hasCreator]-> (person:person)
WHERE tag.name = ?
GROUP BY person, tag
ORDER BY 1*messageCount + 2*replyCount + 10*likeCount DESC
, person.id
LIMIT 100
• A scalar subquery is a subquery that
returns a single value (single row + single
column)

PGX API: execute PGQL queries and print results
• Execute a PGQL query:
• Print the results:
34
resultSet =
pgxSession.queryPgql("SELECT ... FROM myGraph MATCH ...")
// prints all results to standard output
resultSet.print()
// prints the first 100 results to standard output
resultSet.print(100)
resultSet =
pgxGraph.queryPgql("SELECT ... MATCH ...") •
In the Groovy shell, use triple
double quotes so that it’s not
necessary to escape new lines
The FROM clause can be ignored when executing
queries against a PgxGraph (instead of PgxSession)

PGQL API: iterate through results and close result set
• Iterate through the result set - Java iterator way:
• Iterate through the result set - JDBC way:
• Close the result set:
35
for (PgxResult result : resultSet) {
String movie_title = result.getString("m.title");
String movie_budget = result.getFloat("m.budget");
...
}
while (resultSet.next()) {
String movie_title = resultSet.getString("m.title");
String movie_budget = resultSet.getFloat("m.budget");
...
}
resultSet.close()

Prepared statements
• Safeguard applications from query injection
– I.e., we want to prevent applications from using string concatenation:
• Injection attack: firstName = "Franklin", lastName = "Wood' OR '1' = '1"
• Prepared statements also speed up query compilation
– I.e. compile query once and execute it many times but with different bind values
36
String query = "SELECT n.street_address " +
"WHERE (n), n.first_name = '" + firstName + "' AND n.last_name = '" + lastName + "'");
SELECT n.street_address
MATCH (n)
WHERE n.first_name = ? AND n.last_name = ?
Question marks
(?) for bind
variables.

PGQL API: creating and executing prepared statements
– Creating a prepared statement:
– Setting the bind variables
– Executing the prepared statement
• this requires that all bind variables have a value
37
stmnt = pgxSession.preparePgql("SELECT ... WHERE ...")
stmnt.setString(1, "xyz") // first bind variable has index 1
stmnt.setInt(2, 100)
stmnt.setDouble(3, 2898.10)
...
resultSet = stmnt.executeQuery()

PGQL API: prepared statements
• Example: executing a prepared statement twice but with different bind
values
38
stmnt = pgxGraph.preparePgql("SELECT n.street_address " +
"MATCH (n) WHERE n.last_name = ? AND n.income >= ?");
stmnt.setString(1, "Wood");
stmnt.setFloat(2, 50000);
resultSet1 = stmnt.executeQuery();
stmnt.setFloat(2, 100000);
resultSet2 = stmnt.executeQuery();
stmnt.close();
bind values:
[“Wood“, 50000]
bind values:
[“Wood“, 100000]
Closing the statement
also closes the (two)
associated result sets

Date-time data types in Java and mapping to PGQL types
• Using the new Java 8 Date-Time library1
– java.time.*
– New library has five main classes, one for each of the five PGQL date-time types
• Note: types correspond but names differ (e.g. TIMESTAMP <--> LocalDateTime)
39
Data type Result Set API (Java) Prepared Statement API (Java)
DATE LocalDate getDate(..) setDate(LocalDate)
TIME LocalTime getTime(..) setTime(LocalTime)
TIMESTAMP LocalDateTime getTimestamp(..) setTimestamp(LocalDateTime)
TIME WITH TIME ZONE OffsetTime getTimeWithTimezone(..) setTimeWithTimezone(OffsetTime)
TIMESTAMP WITH TIME ZONE OffsetDateTime getTimestampWithTimezone(..) setTimestampWithTimezone(OffsetDateTime)
1 Java SE 8 Date and Time - http://www.oracle.com/technetwork/articles/java/jf14-date-time-2125367.html
alternatively, access a date/time/timestamp as a
java.util.Date through getLegacyDate(..)

Copyright © 2017, Oracle and/or its affiliates. All rights reserved. | 40
Graph Query (PGQL)
In-memory Analyst / PGX
• Extreme performance, especially for analytical queries
• Can combine graph algorithms with graph queries
• Efficient property update through API
(topology update only through Delta Update)
Oracle RDBMS
In-memory Analyst (PGX)
Delta Update
- Synchronizes an in-memory
- graph snapshot with graph
- changes from RDBMS
- Every x seconds/minutes/hours
- or upon request
/* find friends of friends of Clara */
SELECT fof.name
FROM myGraph
MATCH (p:Person) -/:knows{2}/-> (fof:Person)
WHERE p.name = 'Charles'
SELECT fof.name
FROM myGraph
WHERE p.name = 'Thomas'
SELECT fof.name
FROM myGraph
WHERE p.name = 'Clara'
RDBMS (PGQL-to-SQL)
- Schema-less (the graph is stored as triples)
- Efficient property and topology update through
API/SQL
- Can query very large data sets that don’t fit in
the memory of a single machine
Transactional graph query
PGX
RDBMS
Analytical graph query
PGQL in Oracle Spatial and Graph (OSG)

41
Resources

Oracle OpenWorld and Code One 2018
• Oracle‘s flagship conference with expanded developer content
• Graph sessions and demos will include:
– Introduction to Graph Analytics and Graph Cloud Service
– How to Analyze Data Warehouse Data as a Graph
– Anomaly Detection with Graph Analytics
– Build Serverless Big Data and Graph Visualization Web Applications using Spring Data and Core Java (Tom
Sawyer Software)
– Graph Query Language for Navigating Complex Data
– Extend BI/DW by Modeling a Star Schema as a Property Graph in Oracle Database 12.2
– When Graphs Meet Machine Learning
– Analyzing Blockchain and Bitcoin Transaction Data as Graphs
• View full list at tinyurl.com/SpatialGraphOOW18
oracle.com/code-one | October 21-25, San Francisco, CA

Analytics and Data Summit
All Analytics. All Data. No Nonsense.
March 12 – 14, 2019
Formerly called the BIWA Summit with the Spatial and Graph Summit
Same great technical content…new name!
www.AnalyticsandDataSummit.org

Resources
• Oracle Spatial and Graph & Big Data Spatial and Graph on OTN
oracle.com/technetwork/database/options/spatialandgraph
oracle.com/technetwork/database/database-technologies/bigdata-spatialandgraph
– White papers, software downloads, documentation and videos
• Blogs – examples, tips & tricks
blogs.oracle.com/oraclespatial | blogs.oracle.com/bigdataspatialgraph
• Property Graphs 101: How to Get Started with Property Graphs on the Oracle Database –
Arthur Dayton, Vlamis Software https://youtu.be/QSj0zOjOAWI
• YouTube channel: https://www.youtube.com/channel/UCZqBavfLlCuS0il6zNY696w
• Oracle Big Data Lite Virtual Machine - a free sandbox to get started
www.oracle.com/technetwork/database/bigdata-appliance/oracle-bigdatalite-2104726.html
– Hands On Lab included in /opt/oracle/oracle-spatial-graph/ or http://github.com/oracle/BigDataLite/
44

Resources – social media and online communities
• Follow the product team: @SpatialHannes, @JeanIhm, @agodfrin
• Oracle Spatial and Graph SIG user groups (search “Oracle Spatial and
Graph Community”)
45

AskTOM sessions on property graphs
• Next Spatial and Graph session in November
– Topic to be announced – stay tuned
• View recordings, submit feedback, questions,
topic requests, view upcoming session dates and
topics, sign up to get regular updates
46

47
Thanks for attending! See you next month.

PGQL: A Language for Graphs

More Related Content

What's hot (20)

Similar to PGQL: A Language for Graphs (20)

More from Jean Ihm (6)

Recently uploaded (20)

PGQL: A Language for Graphs