A spatial DBMS buyer’s guide

Spatial DBMS Buyer's GuideOTB Research Institute for Housing, Urban and Mobility Studies

2008-10-01 FOSS4G Cape Town

1

A spatial DBMS buyer’s guide

Wim de Haas - RWSWilko Quak - TUDelftMaarten Vermeij - TUDelft

OTB Research Institute for Housing, Urban and Mobility Studies

2008-10-01 2Spatial DBMS Buyer's Guide

Overview

• Short CV• Example: A camera buyer’s guide• Towards a spatial DBMS buyer’s guide• Discussion

OTB Research Institute for Housing, Urban and Mobility Studies

2008-10-01 3Spatial DBMS Buyer's Guide

Background

• 1994: PhD student in spatial DBMS performance• 1995: Wrote spatial extension to MonetDB (was

fastest in the world back then)• 2000-now: work at GIS-technology department

and test various DBMS for various reasons:• Products: Ingres, Oracle, Oracle topology,

Oracle Point Cloud, Informix, PostGIS, MySQL, ArcSDE, …

• Datasets: Cadastral, Laser Scanning Points, GPS logs, road maintenance data.

OTB Research Institute for Housing, Urban and Mobility Studies

2008-10-01 4Spatial DBMS Buyer's Guide

Example: A camera buyer’s guide

Figure out your

profile

Check out the specs

that matter

Use featurematrix

OTB Research Institute for Housing, Urban and Mobility Studies

2008-10-01 5Spatial DBMS Buyer's Guide

Towards a spatial DBMS buyer’s guide?

• Buyer’s Guide (does not exist)• Feature Matrix:• Last version compares only (SQLServer,

MySQL and PostGIS.• Based on documentation and not verified

OTB Research Institute for Housing, Urban and Mobility Studies

2008-10-01 6Spatial DBMS Buyer's Guide

Let’s write a ‘buyer’s’ guide:

1. Make categories of users:• Done: FOSS4G2007: Server Builders, GIS

Users, Dataset Maintainer, Power User2. Figure out relevant feature for each category:• Todo: See rest of presentation

3. Fill feature matrix:• Continuous JOB: For community?

OTB Research Institute for Housing, Urban and Mobility Studies

2008-10-01 7Spatial DBMS Buyer's Guide

Buyer’s guide Feature list

• Functionality• Performance• Ease of Use• Documentation• Hardware needs• Total Cost of Ownership• Integration in business environment• Scaleability• Open Source vs Closed Source

OTB Research Institute for Housing, Urban and Mobility Studies

2008-10-01 8Spatial DBMS Buyer's Guide

Functionality

• Check the documentation.• Run cross check or test:

Same co-ordinate transformation gives meters of

difference

OTB Research Institute for Housing, Urban and Mobility Studies

2008-10-01 9Spatial DBMS Buyer's Guide

Performance (= benchmarking)

• Not as important as you would expect• In most cases performance is not an issue• Moore’s law helps a lot• There are big differences

between query optimizers

OTB Research Institute for Housing, Urban and Mobility Studies

2008-10-01 10Spatial DBMS Buyer's Guide

Ease of Use

• For 1 day of consultancy to improve performance you can buy a lot of hardware!

• A cumbersome system is not fun:• A self organizing DBMS might save a lot of work:• Adherance to standards saves time (Learn

another dialect of SQL, Please: polygons in WKT)

• Readable documentation helps

OTB Research Institute for Housing, Urban and Mobility Studies

2008-10-01 11Spatial DBMS Buyer's Guide

DocumentationPostGI

S

Oracle

OTB Research Institute for Housing, Urban and Mobility Studies

2008-10-01 12Spatial DBMS Buyer's Guide

Total Cost of Ownership

• Open Source = free• Cost for support• Hard to find support staff

OTB Research Institute for Housing, Urban and Mobility Studies

2008-10-01 13Spatial DBMS Buyer's Guide

Integration

• How well does the DBMS integrate with other products:• SOAP• JDBC• Hibernate• The rest of the GIS stack

OTB Research Institute for Housing, Urban and Mobility Studies

2008-10-01 14Spatial DBMS Buyer's Guide

Scaleability

• Very hard to predict without testing.• Most functions look like this:

There is a breakpoin

t here

There might be another breakpoint

here

OTB Research Institute for Housing, Urban and Mobility Studies

2008-10-01 15Spatial DBMS Buyer's Guide

Open Source or not?• Vendor Lock-in is EVERYWHERE:

Marks from old non-standard sized toilet paper dispenser

OTB Research Institute for Housing, Urban and Mobility Studies

2008-10-01 16Spatial DBMS Buyer's Guide

Discussion