Technical Aspects to GIS in the Cloud Nathan Watermeier State GIS Coordinator State of Nebraska OCIO Michael Schonlau GIS Manager Douglas County 2013 NSGIC Annual Meeting October 30, 2013 NebraskaMAP.gov
Douglas County Nebraska Moved 2-1/2 years ago to Cloud Physical hardware for geodatabases on site Cloud-hosted applications and geodatabases 13 production web apps running on EC2 servers Replicated Enterprise Geodatabase (SDE & SQL) on EC2 server Various web services and GP tools 100+ REST services (including cached) Always testing new versions of software Hoping to make better use of scripts
Initial Assessment How are we going to deal with Big Data? What is our infrastructure situation? State level data centers Should we be a Big Data customer? Do we have viable internal options? Do we have the authority and support to move to the cloud? What about security? Longer term goals for server environment?
Cloud Setup Amazon Web Services (AWS) account aws.amazon.com credit card required (PO s possible) Remote admin tools Windows - Remote Desktop Mac - CoRD (open source) Esri or other license(s) EDN ArcGIS Server Standard or Advanced SDE (ArcGIS Server Basic) Access to Esri Amazon Machine Images (AMI) (contact customer service)
Costs Amazon Web Service (AWS) Mostly a la carte pricing Pricing varies by region Amazon Storage Platform (S3) http://aws.amazon.com/s3/pricing/ Based on quantity Reduced redundancy available at lower cost 99.99% durability Separate data transfer pricing
Amazon Elastic Compute Cloud (Virtual Servers) EC2 http://aws.amazon.com/ec2/pricing/ Instance types On Demand - flat hourly rate Reserved - upfront fee, lower hourly rate Spot - bid on unused EC2 space, variable Data transfer - monthly i/o EBS volumes - per Gb, per month Elastic IP s Monitoring (CloudWatch) Elastic load balancers
Douglas County Cost Example
Other Costs Part of your long-term strategy Personnel Time Software Server (ArcGIS Server, # cores) Desktop (ArcGIS) Other software for processing and data sharing
State Level Perspective State is currently operating in our own VM environment The Amazon Cloud scenario proved to be scalable (machine time and storage) with image caching Going from one county to 93 counties of imagery data Ability to cache a statewide image in 80 hours (cache size 57.7 Gb (14,886,215 tiles)) Data storage costs are half the cost of the state Estimated total cost savings are around 21% savings per year considering the Cloud perspective The state recently expanded our SAN, hardware and VM capacity
Advantages Small IT shops - spending more time on GIS vs server admin Flexibility Easy test instances Administer from anywhere Elastic IP s and scalable storage No VPN - just remote desktop tools Reliability Security Redundancy Performance Cost - when you consider all costs
Disadvantages Less control over physical environment IT resistance Costs - if you don t manage them Sensitivity to housing secure data in cloud Fear of the unknown
What we learned The cloud is not a mystery This process provided perspective on technical and management considerations for big data How you manage your workflows and data between the various cloud services had a play in our costs Think about audience of product deliverables and interfaces Costs, performance, software, security, scalability, and adaptability It s hard to plan long-term for GIS - stay flexible Make smart, strategic investments - It s better to be agile
Thank You Nathan Watermeier State GIS Coordinator Nathan.Watermeier@Nebraska.gov Mike Schonlau GIS Manager MSchonlau@douglascounty-ne.gov