Transcript Cloud GIS - Geospatial World Forum

Desktop GIS, Web GIS, is Cloud the next Step?
Prabuddha Ghosh
20th Jan 2011
GeoSpatialWorld2011 Developers Forum
Timeline of GIS Platforms
•The What and Why of Spacial Cloud Computing
(SC2)
•Challenges and solutions of SC2 on AppEngine
•CloudGIS
What is the Cloud?
NIST Definition :
http://csrc.nist.gov/groups/SNS/cloudcomputing/index.html
 5 Characteristics of Cloud Computing
 3 Service Models
 4 Deployment Models
Cloud Characteristics
 On-demand self-service
 Broad network access
 Resource pooling
 Rapid elasticity
 Measured Service
Cloud Service Models
 Cloud Software as a Service (SaaS).
 Cloud Platform as a Service (PaaS).
 Cloud Infrastructure as a Service (IaaS).
Cloud Deployment Models
 Private cloud.
 Community cloud.
 Public cloud.
 Hybrid cloud.
Comparing Cloud & Data Centers
(Source Communications of the ACM)
What is Spatial Cloud Computing
SC2?
 Cloud Computing with Geography added
 90% of business data has a Geographical component
Why SC2 for Clients
 On Demand Scalability
 Low upfront Cost
 No IT costs
 High security
 Latest software
 Opex instead of Capex
Why SC2 for IGSV
 Sell to new categories of customers
 Concentrate on core competency
 Monthly revenues
 Fast to market
 Support massive scaling with low upfront cost
Costs
Traditional
GIS
Open Source
GIS
Open Source
Cloud GIS
Managed
Cloud GIS
License Cost
Yes
No
No
No
Hardware Cost
Yes
Yes
Paid Monthly
direct to Cloud
Provider
(Google)
Included in
Monthly fee
Software
Customization
Cost
Yes
Yes
Yes
Included in
Monthly fee
Data Cost
Yes
Yes
Yes
Included in
Monthly fee
Maintenance
Cost
Yearly
Yearly
Yearly
Included in
Monthly fee
Growth of Cloud Computing
Challenges and Solutions of SC2 on AppEngine
Technology Stack Used
Software type
User Interface Library
Software
Mapping toolkit
Google Web Toolkit (GWT) Version
2.1
Google Maps API Premier
Spatial Indexing toolkit
JavaGeoModel
KML read/write library
JavaApiForKML
Charting Libraries
Google Charts
Cloud Computing Platform
Google AppEngine for Java
Cloud Database
Google Datastore
What is Appengine?
 Google Cloud Platform
 Python & Java
 Self Service Administration
 Integration with Google APIs (GALGWT)
AppEngine Architecture
Dashboard
What is the DataStore
 Not an RDBMS.
 Scalable
 NoSQL based pseudo HashMap
 JDO/JPA support
DataStore Viewer
Other Supporting components
 GWT
 GALGWT ( maps, charts, reports)
http://code.google.com/p/gwt-google-apis/
 JavaApiForKML
http://code.google.com/p/javaapiforkml/
 Apache Commons FileUploading etc
 JTS (For spatial analysis)
 Any Other Java component on whitelist
http://code.google.com/appengine/docs/java/jrewhitelist.
html
Structure of an Appengine Application
 UI- GWT/GXT/SmartGWT (Flex is an option)
/Client Directory
 Middleware (Java based POJOs) /server directory
 Database – Google Bigtable based Datastore
 .gwt.xml file Module descriptor
 Web.xml, appengine-generated.xml , datastore-
indexes.xml, local_db.bin
 Main html Page
Tools for working with AppEngine
 Google plug-in for Eclipse
 AppEngine Datastore
 Appspot
 SVN/ Subclipse/ Googlecode
 Appcfg.sh+ Ant
 Local Admin (ah_admin)
Major Challenges
 Appengine DataStore does not have Spatial Indexing
 Loading Data to Appstore from any machine in the
world as Java Appengine does not have good bulk
upload tools
 Supporting GIS functionalities without reinventing the
wheel
Solutions
 We can use JavaGeoModel an existing open source
project which supports creation of spatial indexes
using Geohashing feature.
 Use Apache Commons Framework and JavaAPIforKML
for handling the data upload task
 Use Java Topology Suite for GIS processing
What is Geohashing?
Store GeoHash in Datastore
RMSI in the Cloud
Where we are?
 POC created on Google Appengine platform
 Work started for a telecom sector client using this
platform
Functionalities available now
 Can load KML data to Appengine from any computer
 Can do proximity searches using the DataStore Geohashes
 Display the same using Google Maps and make available
Google Maps Based Tools
 Can do Attribute Based Search, LatLong Search, Geocoded
Search, Administrative Boundary Based Drill Down Search.
Screenshots/Demo
Screenshots/Demo
Screenshots/Demo
Screenshots/Demo
Screenshots/Demo
Screenshots/Demo
Screenshots/Demo
Screenshots/Demo
Screenshots/Demo
Screenshots/Demo
Screenshots/Demo
Screenshots/Demo
Screenshots/Demo
Screenshots/Demo
Future Work
 Support formats besides KML
 Integrate Administration with AppEngine Admin





Panel
Create Pay as you go Billing (using AppEngines
existing billing engine)
Expose more JTS functionalities
Utilize TaskQueues for running background reports
Port to VMForce when VMForce is in General
Availability (port to Spring in preparation)
Support user created data
Adding Data via user clicks
In Conclusion
 We saw what is SC2 and why it makes sense
 We talked about AppEngine and how to use
AppEngine for a SC2 implementation
 And we had a look at an implementation of an SC2
application for Tower Location Planning
Something to Think About
 Cloud Computing in India to grow from a from a $50
million industry in 2009 to a $15 billion industry by
2013. (Source Communications of the ACM). How
much will be Cloud GIS??
Questions?
Supplementary Slides
Costs comparison
 Traditional Web GIS – Total Costs 200K, Development
Cost 55K, 1 year Maintenance Cost 30K, License Cost
100K, Hardware Cost 20k, 1 Year COTS AMC 20K
 IAAS- Total Costs 150K. No Hardware Cost. Lower
maintenance Cost balances monthly usage charges
 PAAS- Total Costs 100K. 80K development cost and
20K Maintenance over one year period.
Example Total Cost Ownership
Traditional GIS with
Commercial Software
(Rs)
IAAS with same
Commercial SW(Rs)
PAAS with Open Source
(Rs)
Hardware
Yearly Hardware AMC
1800000
360000
0
0
0
0
COTS License
COTS AMC
Development
Yearly Customized Software
Maintenance
4500000
900000
2700000
540000
4500000
900000
2700000
540000
0
0
3600000
0
Monthly Cloud Usage
Charges
Monthly IT costs
Monthly Cloud SW
Support
TCO for 3 years
RMSI Share
Upfront Cost
0
45000
45000
45000
0
0
16020000
27%
56.17977528
16380000
46%
43.95604396
0
9000090000
8460000
81%
42.55319149
TCO for Project X