I have two absolutely identically configurations, this is a cluster of 2xHP DL-380 G5 connected with gigabit link
Created 2 absolutely identical logical volumes.
First logical volume i've formated in EXT3 on both nodes and glued it with Glusterfs, on the base of second one i created the /dev/drbd0 device and deployed lvm volume group on it (with the help of this: https://bugzilla.redhat.com/show_bug.cgi?id=530881).
All absolutely the same, but on DRBD disc write operations throughput of KVM guests is approximately 3Mb/sec and on GlusterFS-based storage (with the same Mandirva 2010 as guest OS) - 23 Mb/sec...
What is is your choice in this situation?..
Created 2 absolutely identical logical volumes.
First logical volume i've formated in EXT3 on both nodes and glued it with Glusterfs, on the base of second one i created the /dev/drbd0 device and deployed lvm volume group on it (with the help of this: https://bugzilla.redhat.com/show_bug.cgi?id=530881).
All absolutely the same, but on DRBD disc write operations throughput of KVM guests is approximately 3Mb/sec and on GlusterFS-based storage (with the same Mandirva 2010 as guest OS) - 23 Mb/sec...
What is is your choice in this situation?..
Last edited: