Hi Ken,
Sorry for not getting back earlier. Wrapping up 2009 proved a bit more cumbersome than expected :(
All the best for 2010!
We have just concluded our first year of 'collaboration infrastructure', COIN in short. As you may remember, in the first half we investigated what an infrastructure for collaboration should look like from an architectural viewpoint and investigated the possibilities for connecting generic collab services to such an infrastructure based on open standards. The ability to 'domesticate' was one of the things that poped up as a very important success factor for making such an infrastructure usable.
The report was published in Q3 and is now available at <
http://www.surfnet.nl/Documents/indi-2009-07-020%20(Report%20Collaboration%20Infrastructure).pdf>.
Based on this - and our the discussions we had earlier on how a VO platform should look like - we set about to try and see how this works out in the real world. We decided to focus on three areas:
1- Integrate generic, domesticated, collab services 'hosted' by SURFnet through the infrastucture. Software modified or patched to act domesticated includes: Adobe Connect; Alfresco; Foodle; Filesender; Confluence; WSO2 mashup server; OpenFire; Drupal; KnowledgeTree and Limesurvey
2- Integrating generic, domesticated, third party collab services through the infrastucture Services domesticated (through available api's) including: Google Apps; MyExperiment.org; Twitter; PubMed
For both of these domains, we looked at either directly accessing a federated service, as well as using techniques for accessing data API's provided by the services for which the infrastucture would then access the services on behalf of the federated user (oAuth, WS security).
3- Look at the ability to create a coherent view of all the apps available for a user in case these services are distributed over the VO, institution and a third party domain. For this we used OpenSocial portal tech.
You may know OpenSocial (
www.opensocial.org) already as it is for example the basis for iGoogle. We implemented a so called OpenSocial container (Shindig) which we then domesticated. Interesting thing on OpenSocial is that, apart from the standardized widget/gadget spec for doing GUI stuff (much like JSR168), OpenSocial also has a standized API for what they call 'Social data': a persons attributes, his group relations and his activities. By provisioning the container using federation arttributes and group relations, this API can now be used to get this kind of data into applications *without* the need to adapt backend services for domestication. Just protecting a OpenSocial service with standard stuff like Shib of SimpleSaml is good enough to do access control, the rest of the user and group provisioning is handled using the OpenSocial API standard.
We found this works very well. We could for example have a user log into the portal using his SURFfederatie account, and then seamlessly login to the services that present their widgets in the portal. As portal and backend service have the same context on the user, we we able to e.g. present only the document sharing directories of the groups the user belongs to, just provide the twitter feeds preselected by his group, use the group as the adressees for the Foodles, etc.
Finally we also tested the ability to use the widgets/gadgets in a generic OpenSocial environment like iGoogle, and this also works in most cases, although a user may be presented a federated login screen, depending on the last time he/she loged in into another service.
In general OpenSocial Gadgets seems like an interesting candidate to do the GUI stuff for a VO, while the data API may provide the standards based API for handling people and groups. Apart form the current usage in Social Networks (100Million users monthly), we note a steady uptake in the educational world. Upcoming Sakai 3 will have a Shindig container in it's backend and so will Confluence 3.1. Also the Grid community is picking this up; see
http://www.collab-ogce.org/ogce/index.php/Main_Page; OGCE Gadget Container.
We are now working on getting a wiki up and running that presents all the results of the past year, including source code for the domestication stuff we did. A labs environemtn featuring a live environment for experimenting with this kind of technology will be in place in 2010. This will partially feature some of the stuff of this project,
In the mean time some of my coworkers have rolled out Grouper as a service spanning group management service for all SURFnet services. We added a userfriendly, federation enabled enduser GUI for which the source code will become available shortly (possibly we present this at the upcoming I2 Spring meeting). We have set up a very basic roles and right management as well. We are now working on domesticating all current SURFnet services, where applicable; a process that will probably take until 2011.
In 2010 COIN has become a programm within SURFnet, that will focus on a number of areas, but is till a bit fuzzy. In general COIN will focus on providing the infrastucture for generic collaboration servcies, while anothe project (GIGAport), will work on e-science applications. We have decided that both programms will try as much as possible ato adhere to the architecture as presented in the report mentioned earlier. For COIN we will probably look at domestication of institutional services, provisioning and deprovisioning in identity federations, bringing some of the stuff we developed in 2009 to production level (including a domesticated version of Java Shindig, which could perhaps fit nicely into CoManage, to add support for social networking), pilot a VO setup where both identity and group management are externalized, pilot collaboration on large data mashups, creating a 'surfnetlabs' environment for experimenting with domesticated services for both institutions as well as third party service providers.
Finally SURFnet will participate in Geant JRA3-T2 (VO's), though my role in that project is rather limited.
I will present some of the results of 2009 at the upcoming TNC, if my paper gets accepted, and I'm also thinking about coming over to Washington for the I2 Spring meeting. Will you visit the next TF-EMC2 by the way?
I look forward to Skyping, preferably at some point towards the end of a week (Thursdays or Fridays)
Kind regards,
Niels
-
On 12/21/2009 06:09 AM, Ken Klingenstein wrote:
Hi Niels.
It's been a long time since we chatted. I hope things are well in Surfnet. At least the Dutch are in the World Cup with a reasonable draw to boot.
It would be good to chat, likely via Skype, in the next few weeks. I'd like to find out the status of collaboration management work in Surfnet, talk about some stuff happening here, etc. Want to suggest some times and dates that might be good? Thanks.
Ken
--
Niels van Dijk
Advanced Services
T: +31 302 305 337 / M: +31 651 347 657
SURFnet - PO Box 19035 - NL-3501 DA Utrecht - The Netherlands -
http://www.surfnet.nlSURFnet - We make innovation work
