JA-SIG, uPortal, SAKAI and CREE

The afternoon started with some worthy, but not very interesting (to me) presentations on the state of JA-SIG generally, and current progress with both SAKAI (the JA-SIG E-learning framework) and uPortal.

There have obviously been some real issues with different versions of uPortal, but it now looks like code developed for v3, will actually be folded back into v2.x (and possibly renamed uPortal NG) – basically there is some housekeeping to do.

Also, they are expecting to do a wide consultation on where uPortal should go next – important that we are aware of this, and understand where SunGard SCT (who produce Luminis) see this going.

I can’t get a handle on the SAKAI stuff – until now it had seemed like it was all coding and no product. However, apparently there are some really big installs in the states now (e.g. 90,000 users at Indiana)

There are apparently a couple of sites (Cambridge and Hull) in the UK seriously looking at this. It was interesting to chat to some of the team from Hull about this – in the end this seemed to be a technology driven thing (they have invested heavily in Java/Webservices/XML skills, and want systems to be based around these technologies).

On a slightly more interesting note Ian Dolphin spoke very briefly about CREE. This was a project looking at how users want to ‘search’ in various environments – including institutional portals.

There was some technical work done showing how various search tools can be integrated using JSR 168 and WSRP (the former being easier to do, the latter proving to be better in this particular context). There are some oddities here – Ian argued this was a ‘thin’ portal implementation, but from a Library point of view it seems ‘thick’ – unless you immediately go from (say) the number of results to the native interface. From my point of view, it would be more efficient to code a single channel for MetaLib, and then add parameters so that the channel can be used to search any MetaLib compliant database (either one at a time, or several at once). This keeps all the ‘hard’ configuration work and metasearch issues in the federated search engine, rather than trying to solve these again within the portal.

CREE also looked at what users wanted, and found that generally users like to have the search tools in context, and especially seemed to appreciate subject specific portals within VLEs – this is definitely something for us to consider in terms of integrating MetaLib into our portal and/or VLE.

Load testing and performance monitoring

A short, but interesting, session on how this had been done at Edinburgh. The best thing was seeing the tools they have used – definitely some stuff worth following up here:

• Load testing – Webserver Stress Tool – Enterprise Edition – v7.x (£600 for site license) by a German company called Paessler.
• Database monitoring – Spotlight on Oracle by Quest (also available for other databases and other systems – e.g. Exchange – something I’ll mention to various people back at base)
• Usage – wusage, looked like it gave a very good set of information about the usage of various websites and applications.

Portals and Content Management Systems

The last sessions of the day, and the main reason I was here.

Firstly Hull presented on their use of HyperContent. They chose this on a number of criteria, but the 3 main pre-requisites were that it would work with Ingres database, it was Java based and that it was free. Hypercontent stores content as XML and uses XSL to ‘style’ it (although not on the fly – this done when the page is ‘published’).

Hull saw CMS as a companion project to their portal project – the portal is just a framework, and requires content to make is useful or interesting. However, they are going beyond just web based content for the portal – they are now doing some of their Public Website with the CMS, and also planning to do their prospectus (presumably printed and online) with the CMS in 2007. They highlighted that ‘ownership’ of content is an issue, and not one that they have solved (and perhaps there isn’t a ‘solution’)

Nottingham have the same setup as us (Luminis as portal, and Luminis Content Management System – LCMS), so I was listening avidly to this, and rather hogged the question/answer session at the end (oops).

The key issues/selling points for them were:

Ease of use for editors – esp. in context editing
Allows you to concentrate on the content rather than the technology
Helps ensure overall quality and consistency of layout and navigation
Ensure accessibility
Introduce formal editorial practicies – workflow
Maximising databse driven content
Standardised navigational structure for all websites
Prefabricated website – merely awaiting content – default pages and templates pre-defined, option pages available within the model website structure

To get Schools (departments) going, they provide support as follows:

Work with Web Team to develop visual ID
Identify Site Manager and Key Contacts
Identify Content requirements with reference to the model – needs and resources
Develop content and enter via ‘Content Collection Tool’
Indentify Content Editors and Content Managers
Typically the process for a single school takes 3-6 months

LCMS is based on Documentum, and it was good to get an overview about what ‘bits’ of Documentum were included – these were:

Content Server (metadata and file system)
WDK – Web Development Kit
Documentum Administrator

These are complimented by the SunGard SCT developed ‘Luminis Site Studio’

This means that with LCMS you get a fully functioning EMC Documentum Content Server including
Versioning
Workflow
Lifecycles
SCT Site Studio
Documentum Administrator
Support from Sungard SCT
More functionality with each version increment

What you don’t get is
Publishing more than 1 version of content at any one time
Documentum licensed as an ECM solution
Direct access to Documentum support
Documentum Webtop/Desktop client

However, it was clear from the comments from the SunGard SCT representative present that these were not inherent limitations – and that additional components could added – and they seem very interested in growing this area, and perhaps even getting into Enterprise Content Management for HE. Nottingham were certainly hoping to see some of the Documentum ready made portlets appearing in Luminis portal in the future.

The actual integration between LCMS and Luminis portal seemed a bit poor. There is something called ‘Automated Luminis Channel Creation’ (ALCC) – which allows you to publish ‘role’ specific information in the portal – however, the functionality here seemed limited, and possibly not that useful.
The other ‘integration’ is an Inbox channel with SSO to LCMS – this is a portal channel which shows outstanding tasks/workflow etc. from LCMS

Nottingham haven’t yet implemented either of these integrations.

In a reverse of the likely approach at RHUL, Nottingham have so far concentrated on externally facing content. They are considering the Intranet now, but feel that much of the intranet content is already in the portal (Module information, SSO to VLEs, Exam timetables, Reading lists etc)

Finally Nottingham said that it takes them 3-6 months to migrate a school website (elapsed time), depending on the staff in the school. However, they can do more than one school at a time (currently working on the outstanding 27 schools – and don’t seem to expect this will affect the overall timescales – 6 months)