ApacheCon Amsterdam 2009; first impressions

This is the review of my first day at the ApacheCon in Amsterdam. The Apache Software Foundation is to the day exactly ten years old; I know the ASF from the bginning, but this is the first conference I visit. ASF today covers more than 60 top level projects and close to 2000 commiters.

I am impressed by the current activities of the ASF. Besides the well-known apache projects httpd, Tomcat, Lucene and JackRabbit, the trend is going into programming in the large. Indeed, most of the projects adresses the java developers; but in contrast to former activities, the technolgies more and more move to services and applications. Solr - the out-of-the-box search engine build on top of lucene - seems to become more prominent that Lucene; not because the API of lucene is bad (it isn't - its really great stuff put into a very simple API), but because with solr this state-of-the-art open-source search engine becomes usable/pluggable by other frontend technologies.

Erik Hatchers session about lucene case studies showed this clearly; With Solr used as a black box, there are demands to widen up the capabilities in the frontend and in the backend: Solritas for example can do the response rendering wich erases the efforts for a separate frontend layer ontop of solr. In the backend, Lucene is moving from a search engine to a database, carrying all the data you need to display your search results. The current developments covers this features, Michael Busch gave us a brief introduction into the details of Payloads (and how to score them in the results), as well as TrieRange and TokenSream.

This trend of aggregating service components using communication API's is only one part; the big picture of whats coming in the near future gave us Raghu Ramakrishnan with his brief glimpses into Yahoos cloud computing. If you ever thought that cloud computing won't touch you: you're wrong. Cloud computing is the natural evolution of programming for the internet. Googles paper about Map/Reduce (yes, that happend neraly 5 years ago in 2004!) is the key technology behind Apaches widely-used Hadoop. There are many applications build upon Hadoop; one of them is Mahout, an apache project about machine learning. Mahout - as Grant Ingersoll teached us - aims to cover many of the topics you can get in contact with: text classification and clustering, spam analysis, user click and amalysis tracking. Interesting: Even labelling (of text) can be done by machines (Grant: “thats boring for people”). Seems that we need new instruments for collaborative filtering, because manual tagging becomes old-fashioned

One of my favorite sessions was the sling presentation by Betrand Delacretaz. Sling is some kind of best-of-breed; its build upon JackRabbit, moving it to an intelligent, fromnt-end driven web server. The powerful examples where fun (even tough they hadn't worked partially). It seems that Sling coveres some of the RoR rules (convention over configuration). Having a JCR (JackRabbit) Repository which can be used with webdav as well as with a webfrontend is fun!

To be continued…