Hadoop vs Ceph and GlusterFS

[Kurt Moesky]

Hi guys,

I am working on a write-up of Hadoop, Ceph and GlusterFS and was wondering
if you could chime in with some benefits of Hadoop over the other two?

I know Hadoop is widely used by the likes of Yahoo and Facebook.

Are there benefits in scaling, management (I like the Ambari interface) etc?

Thanks.

[Charles Earl]

Would it not be better to compare HDFS as the others are distributed file systems?
Charles

Post by Kurt Moesky
Hi guys,
I am working on a write-up of Hadoop, Ceph and GlusterFS and was wondering if you could chime in with some benefits of Hadoop over the other two?
I know Hadoop is widely used by the likes of Yahoo and Facebook.
Are there benefits in scaling, management (I like the Ambari interface) etc?
Thanks.

[Kurt Moesky]

Hi Charles,

That is actually what we're doing, comparing the Hadoop file system to Ceph
and GlusterFS. Just looking for some input from the field as that what you
experts see as the strengths of HDFS over Ceph and GlusterFS.

Thanks,
Kurt

Post by Charles Earl
Would it not be better to compare HDFS as the others are distributed file systems?
Charles

Post by Kurt Moesky
Hi guys,
I am working on a write-up of Hadoop, Ceph and GlusterFS and was

wondering if you could chime in with some benefits of Hadoop over the other
two?

Post by Kurt Moesky
I know Hadoop is widely used by the likes of Yahoo and Facebook.
Are there benefits in scaling, management (I like the Ambari interface)

etc?

Post by Kurt Moesky
Thanks.

[Charles Woerner]

May i also suggest benchmarking against qfs (the Quantcast FileSystem, formerly kosmosfs)?

https://github.com/quantcast/qfs

It's a high performance native (c++) drop in replacement for hdfs and works extremely well with hadoop. It ships with hdfs java api bindings as well as native bindings for a number of dynamic languages, including python.

One of the main strengths of qfs over the other filesystems you are evaluating is that it makes use of configurable Reed-Solomon encoding to improve the storage-to-fault tolerance relationship while maintaining best of class performance. It also supports appends.

With respect to setup, it's
very easy to build and install.

(I work for Quantcast and we use qfs exclusively for these reasons.)

Sent from my iPhone

Post by Kurt Moesky
Hi Charles,
That is actually what we're doing, comparing the Hadoop file system to Ceph and GlusterFS. Just looking for some input from the field as that what you experts see as the strengths of HDFS over Ceph and GlusterFS.
Thanks,
Kurt

Post by Charles Earl
Would it not be better to compare HDFS as the others are distributed file systems?
Charles

Post by Kurt Moesky
Hi guys,
I am working on a write-up of Hadoop, Ceph and GlusterFS and was wondering if you could chime in with some benefits of Hadoop over the other two?
I know Hadoop is widely used by the likes of Yahoo and Facebook.
Are there benefits in scaling, management (I like the Ambari interface) etc?
Thanks.

[Jiayu Ji]

I am not very familiar with Ceph and GlusterFS, but I know they are
centralized file systems. In this kinds of FS, compute nodes and the
storage nodes are separated. If the size of your data increases, the
network may eventually become the bottleneck.

Hadoop is a framework includes storage (HDFS) and computation (MapReduce).
It aims to bring the computation power to the storage node. In this case,
it assigns tasks to where the data is stored due to its awareness of the
data locality. Also, if the size of data increases, you can add more nodes
to the cluster. By doing that, you achieve almost linear scalability.

Post by Kurt Moesky
Hi Charles,
That is actually what we're doing, comparing the Hadoop file system to
Ceph and GlusterFS. Just looking for some input from the field as that what
you experts see as the strengths of HDFS over Ceph and GlusterFS.
Thanks,
Kurt

Post by Charles Earl
Would it not be better to compare HDFS as the others are distributed file systems?
Charles

Post by Kurt Moesky
Hi guys,
I am working on a write-up of Hadoop, Ceph and GlusterFS and was

wondering if you could chime in with some benefits of Hadoop over the other
two?

Post by Kurt Moesky
I know Hadoop is widely used by the likes of Yahoo and Facebook.
Are there benefits in scaling, management (I like the Ambari interface)

etc?

Post by Kurt Moesky
Thanks.

[Chris Embree]

Ceph and glusterfs are NOT centralized files systems. Glusterfs can be
used with Hadoop map reduce, but it requires a special plug in, and hdfs 2
can be ha, so it's probably not worth switching. Ymmv.

Post by Jiayu Ji
I am not very familiar with Ceph and GlusterFS, but I know they are
centralized file systems. In this kinds of FS, compute nodes and the
storage nodes are separated. If the size of your data increases, the
network may eventually become the bottleneck.
Hadoop is a framework includes storage (HDFS) and computation (MapReduce).
It aims to bring the computation power to the storage node. In this case,
it assigns tasks to where the data is stored due to its awareness of the
data locality. Also, if the size of data increases, you can add more nodes
to the cluster. By doing that, you achieve almost linear scalability.

Post by Kurt Moesky
Hi Charles,
That is actually what we're doing, comparing the Hadoop file system to
Ceph and GlusterFS. Just looking for some input from the field as that what
you experts see as the strengths of HDFS over Ceph and GlusterFS.
Thanks,
Kurt

Post by Charles Earl
Would it not be better to compare HDFS as the others are distributed file systems?
Charles

Post by Kurt Moesky
Hi guys,
I am working on a write-up of Hadoop, Ceph and GlusterFS and was

wondering if you could chime in with some benefits of Hadoop over the other
two?

Post by Kurt Moesky
I know Hadoop is widely used by the likes of Yahoo and Facebook.
Are there benefits in scaling, management (I like the Ambari

interface) etc?

Post by Kurt Moesky
Thanks.

--
Jiayu (James) Ji,
Cell: [phone]