Home | About | Sematext search-lucene.com search-hadoop.com
 Search Hadoop and all its subprojects:

Switch to Threaded View
MapReduce >> mail # user >> What's the best disk configuration for hadoop? SSD's Raid levels, etc?

Copy link to this message
Re: What's the best disk configuration for hadoop? SSD's Raid levels, etc?
Hi David,
your first point :

the role of thumb is : "one disk per CPU (or per 1.5 to 2 CPU)" in your
case more parrallel IO could be possible with more disks, but
as you wrote, you have less IO bound processing things might change and a
SSD could speed up shuffle & sort phase, but I suggest to do
some benchmarking with "Starfish". This can tell you the difference between
a std. disk an SSD disk.

2.) I would not recommend RAID in a Hadoop box. Hadoop does the
distribution and also the redundancy (like RAID does) but spread across
multiple machines.

3.) You can configure, how many disks are allowed to fail without crashing
a whole node. So you can replace the drive and block redistribution
is done also automaticly as a part of the self healing features of HDFS.

Best wishes

2013/5/11 David Parks <[EMAIL PROTECTED]>

> We’ve got a cluster of 10x 8core/24gb nodes, currently with 1 4TB disk (3
> disk slots max), they chug away ok currently, only slightly IO bound on
> average.****
> ** **
> I’m going to upgrade the disk configuration at some point (we do need more
> space on HDFS) and I’m thinking about what’s best hardware-wise:****
> ** **
> **·         **Would it be wise to use one of the three disk slots for a
> 1TB SSD?  I wouldn’t use it for HDFS, but for map-output and sorting it
> might make a big difference no?****
> **·         **If I put in either 1 or 2 more 4TB disks for HDFS, should I
> RAID-0 them for speed, or will HDFS balance well across multiple partitions
> on its own?****
> **·         **Would anyone suggest 3 4TB disks and a RAID-5 configuration
> to guard against disk replacements over the above options?****
> ** **
> Dave****
> ** **