Hi all,

I'm interested in contributing to Drill; particularly of interested to me is an 
Azure cloud storage plugin; the idea being that a user of the storage plugin 
would be able to address Azure blob storage in a similar way to they might 
address a LocalFileSystem.


This is a model that is similar to that developed for Azure's HDInsight service 
and in EMR, where WASB:// or S3:// schemes allow access to cloud storage in 
Hadoop rather than using the hdfs:// scheme. I realise the implementation as a 
plugin differs, and indeed is cleaner in Drill to my mind (before the project 
commences!). Performance and data locality trade-offs are significant in all of 
these cloud storage services, so concepts like spanning multiple accounts (to 
maximise IOPs) would be built into the plugin.


Reading the contribution guidelines 
(https://cwiki.apache.org/confluence/display/DRILL/Contributing) states I 
should mail here to check this is in line with what others are doing.


Thanks for your time, folks,

Andy

Reply via email to