Benchmarks

Lidar Data Structures - Octree Vs. Kd Tree

Session Type: 
Tech Session
Presenter(s): 
Michael Considine
Trevor Clarke, Opticks

The Arbitrary Storage Of Lidar Data Can Serve To Obscure The 3d Spatial Information Contained Within The Data. This Presentation Will Compare The Benefits And Deficiencies Of The Octree And The Kd Tree Data Structures Both In Visualization Optimizations And Algorithm Data Access In Open Source Opticks. Discussions Will Include Tradeoffs Between Display Optimizations Including Clipping, Panning And Zooming And Algorithm Data Optimizations Including Nearest Neighbor, 3d Plane Fits, Gradient Mapping, Intensity Mapping, Line-Of-Sight Algorithms, And Autonomous Landing Site Determination.

Speaker Bio: 

Michael Considine is a core developer of the open source Opticks project and Ball Aerospace and Technologies

Schedule info

Techniques for distributed high-speed map tile generation using Mapnik & Node.js

Session Type: 
Tech Session
Presenter(s): 
Simon Tokumine, Vizzuality
Javier De La Torre

Map tile generation can be monolithic, hard to scale and optimise. We present a decoupled open source approach that levers asynchronous highspeed V8 server side javascript with the Mapnik mapping toolkit to generate interactive and queriable data layers for your mapping application.

Speaker Bio: 

Simon is a technical lead at Vizzuality where he is busy working on various geospatial projects. He has a particular interest in developing simple, scalable solutions to common mapping problems.

Schedule info

TinyOWS, what's new for the high performance WFS-T server?

Session Type: 
Tech Session
Presenter(s): 
Olivier Courtin, Oslandia

This session is about TinyOWS, the high performance WFS-T server.

TinyOWS operates on top of a spatial database, and provides an OGC WFS-T interface.

This application born in 2007, is now a real alternative to other WFS implementations, especially when seeking simplicity, performance and standard compliance, cf http://www.tinyows.org.

This presentation will focus on the following points:

  • How do we improved _ again _ TinyOWS performance ? An insight on architecture, configuration tips, comparative benchmarks with others WFS OSS tools and more…
  • What new juicy features are available in latest TinyOWS release ? … surprise, surprise …
  • OGC and INSPIRE compliance, what is done and what is left: OGC CITE Tests, WFS 2.0 and INSPIRE Download service
  • TinyOWS and its related ecosystem: QGIS and/or OpenLayers as WFS-T client, MapServer as a WMS companion…
  • Future, roadmap and conclusion
Speaker Bio: 

Olivier is the main developper and maintainer of TinyOWS since 2007.
He's also involved in PostGIS project, as both projects are somewhere tied.

Olivier is with Vincent Picavet owner of Oslandia,
a small company focused on Open Source GIS.

Schedule info

Speed up your GIS server - run GIS software on solid-state drives (SSD)

Session Type: 
Tech Session
Presenter(s): 
Daniel Kastl, Georepublic Japan & Next Co., Ltd.
Seongbong Kim

Solid-State Drives (SSDs) are recently becoming popular storage mediums in laptops, and are praised for their speed and shock resistance. However, there has been some debate over whether or not SSDs are ready for the data center.

GIS software is known for its high demands on server hardware. Especially hard-drives became a bottleneck when serving map tiles, rendering map images or running spatial computations. SSDs seem to be one possibility to increase IO performance, but optimized server hardware is rare and still very expensive.

We contacted a Korean maker specialising in high-performance SSD servers, mainly used to provide video-on-demand services, and we received two of their enterprise servers for testing and benchmarking with FOSS4G software.

In this presentation we present our benchmarking results and explain the strong and weak points by comparing traditional hard-drives and solid-state disks.

By making use of FOSS4G software we test and compare different scenarios such as

  • Rendering of map images
  • Performance of spatial databases
  • Serving of cached map tiles

Hereby we hope to contribute some additional knowledge about how to tune a GIS server and how to improve server performance for Open Source GIS software.

Speaker Bio: 

Daniel Kastl is founder and CEO of Georepublic UG and works in Germany and Japan. He is moderating and promoting the pgRouting community and development. He's an active OSM contributor in Japan and an initial foundation member of the Japanese local chapter.

Schedule info

Scaling PostGIS Queries with Stado

Session Type: 
Tech Session
Presenter(s): 
Jim Mlodgenski, Cirrus Technologies

Stado, formerly know as GridSQL, provides a powerful and flexible analytical environment allowing users to process large amounts of data using a shared-nothing, massively parallel processing (MPP) architecture with PostgreSQL and PostGIS. Data is automatically partitioned across multiple nodes and each node processes its subset of data allowing queries to be distributed across the cluster and run in parallel. This fully open source architecture allows database performance to scale linearly as servers are added to the cluster while appearing as a single database to applications.

This presentation will demonstrate the 10-20x scalability and performance gains of spatial query running in a Stado environment compared to a single PostGIS instance. We will dig into how Stado plans a query capable of spanning multiple PostgreSQL servers and executes across those nodes using the Tiger data set as an example.

Speaker Bio: 

Jim Mlodgenski is a database architect with more than a decade of experience in enterprise database design and application development. As a member of the founding team and the former chief architect of EnterpriseDB, he is a pioneer in database migration technologies. Currently, he is the founder and CEO of Cirrus Technologies and a recognized leader in the PostgreSQL community.

Schedule info

Syndicate content