Oracle Spatial and Graph at OOW 2012 Sessions

Date/Time Title Location Tuesday, Oct 2 JavaOne Session: Leveraging W3C Linked Data for Loosely Parc 55 - 8:30am-9:30am Coupled Application Integrations Embarcadero Conference Session: Oracle Spatial and Graph: Faster, Moscone South - 10:15am-11:15am Bigger, Better 2-D and 3-D Spatial Solutions 306 Conference Session: Integrating Oracle Database with a Social Marriott Marquis – 10:15am-11:15am Network Foothill F Wednesday, Oct 3 Oracle Exadata, Oracle Exalogic, Oracle Exalytics, and Big Westin San 11:45am - 12:45pm Data Solutions in the Public Sector Francisco - City (Features panelists from US Census Bureau, New York City) Conference Session: Mash Up All Things Location with Marriott Marquis – 1:15pm-2:15pm Android, iPad, HTML5 Web Maps, and Spatial Club Room

. Special offer – Get certified on Oracle Spatial onsite! Test Fest @ Oracle PartnerNetwork Exchange Free for OPN Exchange participants

Date/Time Event Location

Monday – Test Fest @ OPN Exchange – Marriott Marquis, Thursday exam Certification Testing onsite for Oracle Spatial and many Juniper Room times available other Oracle technologies (2 hour slots) • Pre-register for an exam slot at http://www.oracle.com/opnexchange/learn/test-fest • Walk-ins available on a first-come, first-served basis • Proof of OPN membership and PearsonVue ID required

Date/Time Demo Location

Monday-Wednesday Spatial Capabilities in Oracle Spatial and Graph Moscone South, Left S-059 Monday-Wednesday Semantic RDF Graph Technologies in Oracle Spatial Moscone South, Left and Graph S-060 Monday-Wednesday Empowering Business to Drive Process Modeling Moscone South, Right S-234 Monday-Wednesday Oracle Event Processing Moscone South, Right S-230 Monday -Thursday Analyze Big Data Moscone North, Upper Lobby – N-013

4 Copyright © 2012, Oracle and/or its affiliates. All rights reserved. Oracle Spatial and Graph: Faster, Bigger, Better 2-D and 3-D Spatial Solutions Jim Steiner, Vice President, Product Management, Oracle Server Technologies

Dan Geringer, Senior Software Development Manager,

Oracle5 Copyright © 2012, Spatial Oracle and/or its affiliates. and All rights reserved. Graph The following is intended to outline our general product direction. It is intended for information purposes only, and may not be incorporated into any contract.

It is not a commitment to deliver any material, code, or functionality, and should not be relied upon in making purchasing decisions. The development, release, and timing of any features or functionality described for Oracle’s products remains at the sole discretion of Oracle.

. Introducing Oracle Spatial and Graph . Goals for Spatial Features in 12c . Vector Performance Acceleration . Parallel GeoRaster and Enhanced Raster Operations . Real World Feature Model for Network Graph

. Parametric Curve Data Type . Extended 3D and Point Cloud data type functionality . Web Feature Server Console . RDF Semantic Graph Analysis with Spatial Functions . Experiences with Exadata – Oracle Spatial and Graph 11g

. Highlights existing graph capabilities in Oracle Spatial – W3C RDF graph since Oracle 10gR2 – Network Data Model graph since Oracle 10gR1

. Addresses increasing market demand for graph database capabilities – Social Network Graph database popularity – Multimodal and integrated transportation, utility and communications networks

“Lines” “Points” “Polygons” Web Services Geocoding (OGC) Routing Oracle Spatial and Graph SPARQL End Point Inferencing

e1 f1 e3 e2 n2 Rasters n1 f2 3D e4 Network Graphs Topologies RDF Semantic Graphs

Spatial Features

. Spatial Data Types and Models

. Spatial Indexes and Analysis

Graph Features

. Network Data Model graph

. W3C RDF Semantic graph

. Dramatically improve performance – Core Spatial Functions – Vector Performance Acceleration – Parallel Raster Operations . Reduce application logic and support real world analysis by moving complex spatial logic into the database – Real world application features for Network Data Model Graph – Server-based Raster Analytics and Image Processing for multi-terabyte cartographic modeling applications – Richer 3D and Point Cloud analysis and visualization; Parametric curves – RDF Semantic integration with Spatial Functions

ANYINTERACT, INSIDE: 20-30x

GEOM DISTANCE: 40X Oracle Database Locator WITHIN DISTANCE: 10X

VALIDATE GEOMETRY: 4X

Join: 50-100x

Touch: 50x Spatial and Graph option Performance Improvements Contains, Overlaps: 50x

Complex masks: 50x

GEOM.relate: 5-10x

DML single insert: 3x Spatial and Graph option Performance Improvements Coordinate System Transformations: 40-50%

General DML operations: 30-50%

• MANY RASTER FUNCTIONS CAN PARALLELIZE • SERIAL OPERATIONS PERFORM UP TO 3X FASTER • SCALES TO OVER 100X OF TIMES FASTER ON HIGHLY PARALLEL SYSTEMS

Virtual Mosaic Support

. Any large collection of georeferenced GeoRaster objects, rectified or non- rectified, from one or more georaster tables . Advanced spatial queries and on-the-fly transformation and mosaicking over it

Mosaic of Landsat Images

NURBS (non-uniform rational B-spline )

• Mathematically precise representation of freeform surfaces and curves which could be exactly reproduced whenever technically needed.

• Used for highway and rail design, and surface models, like those used for aerospace exterior surfaces, and car bodies, etc.

Feature Representation Network Representation

2. Right click on the shape. 3. At the bottom of the submenu select Performance Enhancements and Vector Performance 4. Select “Fill” at the top of the “Format Shape”

Acceleration 5. Select “Picture or Texture fill” from the options. 6. And select “File” under the “Insert from” option. 7. Navigate to the file you want to use and

8. On the “Format” tab, in the Size group, click on

SPATIAL INDEX STATISTICS . New Algorithms to Collect Spatial Index Statistics . Optimized Execution Plans

20-90% REDUCTION IN REDO LOGS . Improved DML Performance . Less Storage Required

OPTIMIZED METADATA QUERIES

. Kernel level caching . Performance gains for DMLs and Spatial function calls . Optimization especially noticeable in workflows with many fast running queries

ADVANCED ALGORITHMS

. Faster more efficient algorithms . Improved memory management

• Raster algebra supported through a set of algebraic expressions and PL/SQL subprograms • Raster algebra operations work on individual raster cells, or pixels • Apply algebraic functions on raster data to derive new results – Generate new raster layer (map) from two or more raster layers – Raster algebra operations to implement sophisticated analytical algorithms, such as a Normalized Difference Vegetation Index (NDVI) , and TCT (Tasseled Cap Transformation) – NDVI helps classify the amount of vegetation in a region, from none to rainforest

. Rectification of georeferenced raw images . Orthorectification of georeferenced raw image with a DEM . Image masking, Image stretching, and Image segmentation

30 Copyright © 2012, Oracle and/or its affiliates. All rights reserved. GeoRaster Advanced and Virtual Mosaic Support Planned 12c GeoRaster Feature . Mosaic can be persisted, or virtual . Virtual (on the fly) mosaic defined by a SQL statement or view . Supports gaps, no data, and overlapping regions . User defined priority for overlapping regions (Date or SQL ORDER BY) . More advanced mosaicking support: – georeferenced raw images – internal reproject/rectification – common point rules (for example, max value, min value, etc...) – simple color balancing – large-scale image append

31 Copyright © 2012, Oracle and/or its affiliates. All rights reserved. Enhancements to the GeoRaster Core Improve performance through parallel processing of all Raster Algebra and Mosaicking procedures

. Support relational Raster Definition Tables as an additional raster data storage . Users can specify default alpha channel and default pyramid level in the metadata . New Bi-quadratic Interpolation Method to interpolate cell value for any point in the grid – Biquadratic Interpolation Method is also added as a new Resampling type in image operations and pyramiding . Added “OTHER” resampling type to indicate an unknown or external pyramid resampling type imported from files . Operation Progress Reporting is added into many new functions . Significant performance improvement of pyramiding with cubic convolution resampling

. Other New Functions and Procedures • getRasterRange: retrieve min and max cell values • getRasterBlockLocator: another new template function • generateSpatialResolutions: supports resolution generation for pyramids • emptyBlocks and recreateDMLtriggers utility functions . GeoRaster Java API Enhancement • Support for ground control point (GCP) storage and manipulation, GCP georeferencing, reprojection, grid interpolations, and getCellValue . New Concurrent Batch Loading and Exporting Tool • GUI to create GDAL-based batch loading and exporting description files (XML)

. Support Non-Uniform Rational B-Spline (NURBS) curve geometries – NURBS curves represent arbitrary, free-form shapes – Control points and knots guide the shape of the curve – Enables complex shapes to be represented with little data . Support for NURBS curves includes the following – Functions to convert WKT/WKB and GML representations into SDO_GEOMETRY – Validation functions . Spatial indexing . Functions and operators approximate NURB curves for computations (implicit densification).

. Implementation conforms to SQL/MM standard – SQL/MM standard for NURBS curves represents splines, polynomial splines, cubic splines, B-splines and Bezier curves. . SDO_GEOMETRY object type supports NURBS curves representation – 2D and 3D Curves supported – SDO_GTYPE 2002 and 2006 for 2D Curve and MultiCurve – SDO_GTYPE 3002 and 3006 for 3D Curve and MultiCurve

. SDOAPI contains an in memory R-Tree for primary filter operations . New SDOAPI functions, enhanced to support the following in-memory operations – Distance – Inside – AnyInteract . New in memory functions support the following geometries types: – Projected 2D and 3D – Geodetic 2D and 3D – NURB curve types

. Similar to Topology Data Model but not limited to geometry based features – Supports logical networks and physical networks in application representation . Data model to manage node and link “features” with their associated “network elements” (Nodes/Links) – Node features: transformers, sub-stations, etc. – Link features: power lines, transit routes, etc. . Consistency between network features and network elements automatically maintained . Feature level analysis – Find the shortest path between two transformers – Find the shortest path between two transformers, but use only a certain wire type

Temporal Modeling/Analysis

. Traffic Patterns – Record historical travel patterns for different classes of roads – Data collected based on time of day and day of the week . NDM can use traffic patterns to compute shortest paths 10 PM – Find shorted path from point A to B with start time of 8 AM 8 AM – Find shortest path from point A to B and reach destination at 5.30PM . Support NAVTEQ Traffic Patterns format out of the box

Multi-Modal Routing

. Each mode (car, bus, rail, bike, etc) modeled as a separate network Train & Bus . Single logical network represents all modes of transportation . Transition nodes where networks meet . NDM APIs can specify the modes to consider . Out of the box support for transit data published by Bus Only transit authorities – GTFS (General Transit Feed Spec) supported

For millions of customers, find closest store within a specified drive time

. Same underlying data for geocoder and road network . Customers geocode as link id and percentage (instead of longitude/latitude) . 5 mile Network Buffer generates all possible paths . Each persisted path includes: – Covered link IDs, nodes ID, and associated costs . Single database query to find closest store and drive Store Location time/distance for each customer (join on link_id) Customer Location

. Provides a full fledged web services framework for network analysis . Enhanced XML API with Network Constraint and Cost Calculator Support . Integrated with Oracle Spatial and Graph Web Service Framework . A Simple PL/SQL Wrapper (http put/get) on top of the XML API . New Driving Direction API – For NDM users with data in Routing Schema – Generate step by step driving directions format for final path

Enhancements

. Traveling Salesman Analysis (TSP) – Optimal multiple-stop route – Open or closed path

Enhancements Left lane . Enhanced Navigation Support only exit – Multi-link restrictions – Highway has entry point, and left Enter highway – Restrictions where there are no physical barriers with no turn restriction on upcoming left hand exit.

Oracle Confidential 3D and Point Cloud Enhancements

47 Copyright © 2012, Oracle and/or its affiliates. All rights reserved. US Army Corps of Engineers is making extensive use of the Point Cloud data type in Oracle Database 12c Spatial and Graph option to store, retrieve, subset and analyze LiDAR data used to model 3D landscapes to support in-theatre, mission-critical DoD operations.

Across a number of instances for the Corps of Engineers and our DoD partners, we are storing more than 150Tb of data including 100Tb of point cloud data (with more on the way). A single instance currently stores more than 1 Trillion points.

Compared to Oracle Spatial and Graph 11g Release 2, the improvements in Oracle Spatial and Graph 12c have dramatically increased scalability for multisession point cloud creation and provide a considerable savings of storage space.

Michael Smith US Army Corp, Remote Sensing Center

. Pyramiding support for PC and TIN data – PL/SQL API added to generate pyramids for existing point clouds – Points can be repeated in the pyramid levels or mutually exclusive – Useful for visualization applications . Contour generation from PC data – Generate linear geometries that connect points with equal elevation values – A polygon can be specified to limit contour generation to a region – The contour generation process is grid-based in x and y. – Krigging performed to assign heights to grids with no point cloud data.

. DISTANCE, ANYINTERACT and INSIDE operations take height into account . LRS functions also support 3D geodetic data . Height for geodetic data is usually supplied as meters – Or it can be any other UNIT type defined by a coordinate system

Length of geodetic segment whose P1 endpoints have different heights P3 . Endpoint with greatest height (P1) is P2 projected onto the sphere of (P2) … and (P3) is generated. . The distance between P2 and P3 is spherical . The distance of P1 to P2 is calculated as the length of side (P1,P2) of triangle (P1,P2,P3).

Ontology-assist Query

Standard Spatial Query

Find all Streams inside a query search window Water Features Table SELECT feature_id Feature_ID Feature_Type Geometry FROM TABLE (SEM_MATCH( 1 Bourne SDO_GEOMETRY( spatial predicate, …) rdf predicate); 2 Influent Stream SDO_GEOMETRY( …) 3 Brook SDO_GEOMETRY( SEM_MATCH combines spatial operators …) (anyinteract, inside, nearest neighbor, 4 Canal SDO_GEOMETRY( …) within_distance, etc..) with ontology predicates.

. Don’t always have quantitative geometry data – Textual Descriptions . Next to Westin St. Francis Hotel . Inside Union Square . How can semantics help? – Apply description logic (reasoning) to spatial relations . Westin St. Francis Hotel is inside Union Square, therefore it is inside San Francisco. . inside is an owl:TransitiveProperty, inside is owl:inverseOf contains, … – Logical reasoning can be faster than spatial computation of geometries . Example: Find all parks inside California – Minimizes spatial computation on large number of geometries (parks)

Web-based Administrative console . Menu driven GUI simplifies registration of spatial layers with Oracle Spatial and Graph’s WFS – Browse existing spatial layers – Eliminates the need for DBA to run PL/SQL scripts to publish spatial layers . Includes tutorial on how to configure and use WFS . Provide sample request and response pages for WFS queries . Can also be used as a client to other WFS servers – In this mode, only browsing options are enabled

• Point, Line and Polygon geometries can all benefit from EHCC • Lines and Polygons, they must be stored inline (less than 4K in size). • Options include: – COMPRESS FOR QUERY LOW – COMPRESS FOR QUERY HIGH – COMPRESS FOR ARCHIVE LOW – COMPRESS FOR ARCHIVE HIGH

Two ways to compress: Create Table As Select Direct Path Inserts

1. Create Table As Select CREATE TABLE edges_compressed COMPRESS FOR QUERY LOW NOLOGGING AS SELECT * FROM edges;

2. Direct Path Inserts (full code example in presentation appendix) -- PL/SQL Example with append_values hint. DECLARE id_tab ID_TAB_TYPE; edge_tab GEOM_TAB_TYPE; BEGIN -- Population of id_tab and edge_tab shown in presentation appendix FORALL i IN edge_tab.first .. edge_tab.last INSERT /*+ append_values */ INTO edge_ql VALUES (id_tab(i), edge_tab(i)); COMMIT;

. Uniform geometries spatial layers have the same number of coordinates in every row. . Some examples: – Point data (x NUMBER, y NUMBER) – Box polygon (lx NUMBER, ly NUMBER, ul NUMBER, uy NUMBER) – Two point line (x1 NUMBER, y1 NUMBER, x2 NUMBER, y2 NUMBER) – Four point polygon (x1 NUMBER, y1 NUMBER, …, x5 NUMBER, y5 NUMBER) . For much higher compression rates, store uniform geometries as a series of NUMBER columns instead of SDO_GEOMETRY

• Create a function based index on uniform geometries to perform spatial queries • The following ANYINTERACT queries were run on a 116 million row table • Query High compression – 18.17x… queries still very fast.

Anyinteract Uncompressed Query Low Query High Archive High Query 3.92x 18.17x 21.57x

10 acre polygon 1.86 sec 2.02 sec 2.7 sec 12.75 sec (487739 rows returned) (1.08x) (1.45x) (6.85x)