- This topic has 1 reply, 1 voice, and was last updated 5 years, 7 months ago by
.
-
Posts
-
-
-
HDFS can function on top of any underlying file system as well as Operating System. And, Linux provides various file system choices, each with caveats which have an impact on HDFS.
In order to speed up reads for files, disable ‘noatime’, if you are mounting disks solely for Hadoop data.
Basically, we have three Linux filesystem options which are popular to choose, such as:
– Ext3
– Ext4
– XFSwell, Yahoo uses the ext3 file system for its Hadoop deployments. However, for many popular Linux OS flavors, ext3 is the default filesystem choice. Although, for the underlying file system HDFS on ext3 makes a safe choice according to a public test on Yahoo’s cluster.
Moreover, we can say, the ext4 is the successor to ext3, because the ext4 has better performance with large files and also introduced delayed allocation of data, that adds a bit more risk with unplanned server outages at the time of decreasing fragmentation and improving performance.
In addition, on comparing with than ext3, XFS offers better disk space utilization and also it has much quicker disk formatting times than ext3. It says it strats quicker with a data node whicle using XFS.
-
- You must be logged in to reply to this topic.
