java - Use Apache Hadoop JAR files or vendor specific? -
java - Use Apache Hadoop JAR files or vendor specific? -
i creating application hadoop should run on distributions of hadoop provided different vendors like: cloudera, mapr, hortonworks, pivotal...etc. application deployed on application servers weblogic, jboss or can deployed on tomcat also. question here is:- suppose version of these vendors utilize same underlying hadoop version hadoop 2.0, should utilize jar files given these vendors or utilize jar files given apache hadoop?
i mean jar files have same classes apache hadoop have name in them blablacdh5.2blabla.jar
, should utilize 1 or 1 apache? can build single version hadoop 2.0 , utilize vendors. can done or have build different flavours of app vendor distributions.
thanks in advance
it depends on how deep hadoop api threading.
if application submits jobs cluster, ok vanilla libraries long stick 1 specific version. if doing advanced stuff , using hadoop internals, may necessary include vendor specific ones.
java hadoop
Comments
Post a Comment