Christopher Warner Studies and thoughts, usually in coherent fashion.

Alfresco review part two

So after talking with Luis Sala yesterday from Alfresco he cleared some of the misconceptions I had about Alfresco and there were some interesting highlights.

1. Alfresco doesn’t necessarily have to be responsible for templating so essentially it can publish static/dynamic contact and you can include that into the overall design of your page. Personally and from experience this leads to problems depending on your web server platform. IE: Apache uses includes and because there is no management system for that; it can get ugly.
2. Dreamweaver can be used for management of items in Alfresco.
3. Content types are created via XML schema and can actually have sub-schemas. So you can embed a content-type in a content-type or as part of creating a complex-content type. Turning this into a mapped object is handled by Hibernate.
4. Alfresco has it’s own caching system but as with anything you’d have to have your own caching platform for any serious operation. Squid/Varnish + Memcache/Akamai whatever.
5. For enterprise solutions he recommended a contract with lets say Mysql or Jboss. In an enterprise situation one probably already has a contract for Mysql if that’s what you are using for your database setup in production. For Jboss, that’s an added cost but that’s if you use Jboss.
6. Alfresco does relationships except it’s not exposed via the user interface. So one would have to manually setup relationships between content-types. This isn’t such a big deal so long as the relationship stuff works decently.
7. It also has something called renditions that will statically publish the piece of content in a specific rendition or file format. So for instance if you want a list of models based on your model content type you can get back xml, html, json and have that published statically. So you can have different ‘renditions’ of content; mobile, print-only version, teasers etc.
8. Auto indexing of content immediately via Lucene.
9. One major thing for me is the lack of events on objects or specific objects. You can work around this via workflow but it would be really nice to see Alfresco and other content management systems employ proper event systems.
10. Alfresco isn’t just document management even though that’s how it’s primarily used and started.  There are other core pieces that make the above possible specifically WCM; so one technically could have the best of both worlds.

Alfresco review part one

So testing Alfresco thus far it looks like it’s a great management system for documents. I’m not sure if it’s such a great system for management of custom content types though. With my experimenting on Twitter I came across Luis Sala who is the Sr. Director of Technology and Business Development and we are going to have a conversation on Wednesday so I can ask some more questions.

So far, it seems like Alfresco is opensource; there are problems with the community edition that require patches. When I installed the Enterprise edition the patches were automatically downloaded and installed. That’s not a very good demo of the community version because it’s broken. To Alfresco’s credit they do state that it’s possible the community version will be broken at times. Unfortunately it was broken at the time I wanted to try Alfresco. So, they may want to take a look at that. I’m sure I could fix it myself but i’m just trying the product and in comparison with Plone, where I just type one command and i’m done. It’s a little bit more involved (it does have a nice GUI install though if you have X available. I haven’t tried on Windows, because lets face it. No one doing anything seriously in the enterprise is using Windows unless they’ve got cash to burn and enjoy downtime).

That said you can create content types via an XML schema (.xsd) and it uses FreeMarker for templating. On the backend it uses Hibernate for Object Relationship Mappings and Tomcat for the actual application server.

Hibernate is developed by RedHat who happens to also develop and own Jboss; Most of the critical components and high availability seem to be left to those specific components. Which seems ok to me. Jboss is obviously Kix.. (kid tested, mother approved) but now that Oracle owns Sun it’ll be interesting to see what happens with Glassfish.

I have a lot of questions though one of them is that Alfresco claims to be the first Opensource ECM. That’s very odd considering the first version came out in 2005. So it’s barely 4 years old and that concerns me. The opensource components it uses are older than itself. Plone has been around since 1998 I believe and before that I’m sure there are other systems. So it’s hardly first anything I am not sure why they say that or even suggest it. Other stacks have existed for a while. Another question is that because the default application stack has many opensource components do we go to Alfresco support or lets say the Apache Foundation? So, if one buys support for Alfresco and there is a problem in Tomcat; do we go to them and they work to get it fixed or do we go straight to Apache?

The default application stack consist of hibernate, javaserver faces for the UI, whatever relational database you are using (from what I can tell Alfresco doesn’t do anything with straight object databases)., Apache Tomcat for the app server which concerns me slightly; and even more after reading this (http://www.javaworld.com/javaworld/jw-01-2008/jw-01-tomcat6.html) Which seems to classify Aflresco simply as a Java Web application and not an enterprise application. Lucene for full featured text search, etc. I’m not a Java programmer by any stretch of the imagination (Over at the Sun Microsystems campus I jokingly chuckled with one of the Java developers there about how verbose it was by handing him a book I called a tutorial on ‘Hello World” in Java).  So i’m gonna tenatively prime my Java skills, it’ll be good practice anyway. Nowadays I don’t have as much time to delve into the full language unless it’s something I am working with and I have never done any Java programming for the web (I’m not exactly sure how many people are running production media websites with Java that aren’t full on applications or POS stores). WebObjects seems like a more cohesive and useful application server for the long haul though. Something you could build a content management framework on. Plus it plugs directly into Eclipse, Xcode and Apple seems to be using it successfully for their online properties for years now. Obviously no one has done that yet and I’m not sure if there are any media organizations using it for a website sort of setup.

Alfresco has a couple of sites using it (fox.com, sesamestreet.org, amnesty.org, activision.com) I’m not sure if it’s strictly for their web properties or  how deep it goes because there aren’t any case studies like that.

So far from what I have tested over the last couple of days Alfresco seems great at managing document content types. So Word documents, PDF documents etc. The idea of a document as a content type seems to be great and it has all the features Plone has with cataloging and indexing of the text. Programming for it seems a little verbose but it’s Java. In Plone if I want to create a new content type I just use paster and it builds out everything I need. From default unit tests to skeleton code; it’s a little more involved with Alfresco.

All that said I’ll be talking to Luis on Wednesday and the rest of the NYC team hopefully and we can get some more questions knocked down. I’m primarily doing this on behalf of New York Magazine but it could be useful to me in other ways, if you have any questions you want answered email me by Tuesday evening. Stay tuned for part two.