AnsweredAssumed Answered

MapR filesystem IO performance issue

Question asked by li_wei on Feb 28, 2012
Latest reply on Feb 28, 2012 by nabeel
I have just install M5 on 2 nodes ,each nodes hava this hardware configuration:
2*6 core
48G mem
2*10Gb NIC
14*15000rpm sas disk,  2 for RedhatLinux OS ,the rest 12 is for MapR,distsetup like :
#disksetup -W 8 /home/disks
# more /home/disks
/dev/sdb
/dev/sdc
/dev/sdd
/dev/sde
/dev/sdf
/dev/sdg
/dev/sdh
/dev/sdi
/dev/sdj
/dev/sdk
/dev/sdl
/dev/sdm

question is :
after created  a volume on these two nodes, I do a dd test through NFS ,the result is bad:
#mount  vip:/mapr /hd  #vip  is virtual IP  for 2 NFS service which be started up in 2 two nodes;
#cd /hd/MapR/mt        #mt is volume mount point
#time dd if=/dev/zero of=2.test bs=32k count=4000000
4000000+0 records in
4000000+0 records out
131072000000 bytes (131 GB) copied, 264.22 seconds, 496 MB/s

!!!just 496MB/s for two nodes,  I had tested DD by EXT3 for each node (do 2 raid5 group for 12 disks in BIOS),and get a 1000MB/s for each node.

what's wrong with it ,how to improve MapR filesystem performance ?

Outcomes