hadoop-mapreduce-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Thomas Demoor <thomas.dem...@amplidata.com>
Subject Re: About building dummy Custom Distributed File system for hadoop.
Date Fri, 31 Oct 2014 08:45:28 GMT
Hi Vikrant,

YARN uses AbstractFileSystem. Filesystem is the legacy interface (Hadoop

However, there is an abstract class DelegateToFileSystem that (trivially)
enables using a legacy FileSystem in YARN  (see f.i. the Wasb class in

So what you should implement depends on if you want to run YARN, legacy
Hadoop or both. Distributed filesystems are hard, nice to see you are

Good luck,

On Thu, Oct 30, 2014 at 6:52 PM, Vikrant Dhimate <vikrantdhimate@gmail.com>

> We are working on B.Tech project and we want to build dummy Custom
> Distributed File system for hadoop.
> In 2.5.1 version there are 2 classes : 1) FileSystem and 2)
> AbstractFileSystem.
> To make our own custom DFS which one of these two classes should we
> implement or extend to our custom filesystem class. Because in earlier
> version there was only FileSystem class.
> Can you please tell us what is difference between FileSystem and
> AbstractFileSystem classes and when to use each one sand which one should
> we
> implement in our case?
> Thanks,
> Vikrant

View raw message