OpenGeoPortal Developer Survey January 2013 Summary of Results Seven respondents Majority are using GeoServer with Linux operating system and Apache server with Tomcat servlet, with some exceptions. The Boston Public Library is experimenting with OSGeo Web Tile Service Protocol to serve scanned maps. Restrictions management seems to be a work in progress at most institutions. Some are only serving public data at the moment, while others are using Shibboleth and/or a university authentication server. MIT is storing public and restricted data on separate Geoserver instances. Stanford is re-developing OGP in Ruby on Rails, in part to make it more compatible with Hydra. They are hoping to build a hydra head that incorporates GIS data. Detailed Results Technology stack: Institution OS Servlet Server Software Web Mapping Server DBMS Cornell University Linux TBD Apache GeoServer TBD Stanford Red Hat Linux Tomcat ? GeoServer Harvard Linux Tomcat Geoserver Harvard / Boston Public Library Linux None Apache OSGeo Tile Service None MIT RHEL 5.4 Tomcat 5.5 Apache 2.2.3 Geoserver 2.1.3 Oracle 11.2.0 UW-Madison Variable OSX+Windows at present Tomcat/Jetty Apache GeoServer Postgresql/PostGIS Yale Windows XP/ Linux yes Apache Tomcat have not instantiated yet n/a (uses solr index) PostGIS (PostGres) Oracle 11g with ArcSDE 9.3.1 Additional comments on technology stack: Cornell University: “Our current repository is based on Apache, Tomcat, Java, and MySQL. We still need to determine what we will use for OGP.” Stanford University: “We found that the OGP software was problematic and we have decided not to implement it as is. We conducted a code review and uncovered significant security issues, no clear way to contribute code fixes, no documentation, and no code tests. Since OGP is really just a jsp wrapper around openlayers javascript, we made a new project that is a ruby on rails wrapper around the same openlayers javascript. However, our project has tests, we feel more confident about its security, it has a clear upgrade path, and it fits into the rest of our technology infrastructure. The code is available here: https://github.com/sul-dlss/ogp-rails” Boston Public Library: “OK, so here is the rub. Are you familiar with the OSGeo Web Tile Service Protocol? http://en.wikipedia.org/wiki/Tile_Map_Service . The great thing about it is that if all you need to serve is a scanned map, it does everything that GeoServer will do -- but it works with any http server. Apparently, it is not hard to add OSG Tile layers to openlayers, see http://openlayers.org/dev/examples/tms.html . Therefore, I think that if a layer description submitted to OGP had a tag that defined what sort of map service is to be used for the preview, and if the renderer was able to appropriately create a layer for an OSG Tile service, then, potentially a collection such as the LMC@bpl could make a lot of great maps available without needing to have their own WMS. If this does not make sense please let me know.” UW-Madison: “Final configuration has not been set in stone. Current iteration is GeoServer running on built-in Jetty server on Windows box for data; OGP running in Tomcat on OSX server for interface.” Server hardware specifications: Cornell University: “Currently in the process of migrating to a virtual server, details TBD.” Harvard: Dell PowerEdge R410 RedHat ES 6 Xeon CPU - 8 cores 24 GB RAM SAN storage at central operations (SOC) UW-Madison: “Subject to change in near-term... right now, I believe we're working with dual dual-core processors and 8 Gigs of RAM. Hardware is up for replacement cycle over next 1 year.” Yale: “rackspace cloud server (small specs - just a dev installation): RHEL 6.1/ 512 MB RAM /20GB storage + windows XP workstation/ 3.25 GB RAM/ 150 DB storage” Three responding institutions are using Hydra: Cornell University: “Cornell just recently started using Hydra for a large library project to eventually replace our public catalog interface.” Stanford University: “We are some of the primary developers of hydra and we re-developed OGP in rails in part to make it compatible with our hydra environment. We hope to build a hydra head that incorporates GIS data.” Yale: “Yes, we are using Hydra, but not with OGP (yet). Currently we are planning on hydra for images and digital collections.” Additional information: Institution IT support model Strategy for managing restrictions on accessing data Cornell University Currently have one official IT developer who mostly works on other projects, so we are looking to bring another developer in for the initial installation and configuration. And, although I am not officially IT, I do some programming, database, UI design, and troubleshooting as well. We currently avoid the issue by only hosting open public data. However, we are increasingly acquiring restricted datasets that we would like to serve online to our campus users, so we will need to work that out eventually. Stanford We previously had a single developer but we now plan to have two fulltime developers building out our spatial data infrastructure, migrating our GIS data to a managed environment, and deploying geoportal (ogp-rails). We aren't there yet. Harvard The project team consists of 1 developer, 1 metadata cataloger, and 2 GIS Specialists that georeference scanned maps and ingest data. Access to restricted data is controlled by the university authentication server (PIN). Harvard / Boston Public Library The LMC@BPL has two catalogers and several other employees. For system administration they depend on the Boston Public Library which is not particularly interested in running special servlets. Unknown. Presumed everything they post would be open to the world. MIT I am the only developer involved and work half time supporting OGP and other GIS needs. The server, both hardware and VM, is managed by central IT. UWMadison Technical lead + variable number of studenttalent-level developers. We have one Geoserver instance for public data and one for restricted data. Authentication is managed at the Apache level with Shibboleth. Planning to implement campus Shibboleth protocol for campus single-sign-on authentication at data server tier - until such time as we have more public content to differentiate. Yale (just a dev install) 1 systems/developer, 1 metadata librarian Have not instantiated access restrictions, however most likely would use CAS