I want to set up a small search cluster and I was wondering if raid is
really needed since ES already has replication built in.
Can I use multpile data paths in the ES config pointing to /mnt/disk1 ...
/mnt/disk12 since disk cost is a premium or do you recommend to put the
data on a raid volume ?
I use RAID0 on data folder for speed. If RAID fails, the whole node is
decommissioned, there is one node per server. Replica level must be greater
0 or you will lose data sooner or later.
Multiple data paths will just scatter files, probably over partitions,
equal balancing over data paths is not guaranteed. And without RAID,
scattering is not faster for read/write so you don;t have any advantage.
Jörg
On Wed, May 14, 2014 at 5:13 PM, Ronny Vaningh ronny@guard-it.be wrote:
Hi
I want to set up a small search cluster and I was wondering if raid is
really needed since ES already has replication built in.
Can I use multpile data paths in the ES config pointing to /mnt/disk1 ...
/mnt/disk12 since disk cost is a premium or do you recommend to put the
data on a raid volume ?
Apache, Apache Lucene, Apache Hadoop, Hadoop, HDFS and the yellow elephant
logo are trademarks of the
Apache Software Foundation
in the United States and/or other countries.