Hi, I am planning to set up a multiple node configuration of Lustre with 1 MDS, 3 Clients and 12 OSTs(4 on each OSS). I will have around 4TB of total storage available with all the OSTs put together. As preliminary tests I just want to run some simple scripts that simulate some extreme workloads to get a broad comparison. for eg test it for a large DB and a small DB load and see how the performance differs. This will all be a hands-on experience in understanding lustre. This is what I have planned. 1) there will be 1 MDS with around 150GB of storage. 2) The LOV would be 1 LOV with name lov1 with stripe size of 2MB and stripe count of 2. 3) On each OSS I plan to have 4 OSTs and all OSTs on every OSS will be associated with the same LOV i.e. lov1. 4) The clients will each be mounted on a different node (Machine) and each will be associated with the mds and lov1. Am I correct in my configuration idea? including the size of the mds. Also, on each client I plan to have the mount point as /mnt/lustre and nid of ''*'' so that its a generic client. Is that correct? Thanks -------------- next part -------------- An HTML attachment was scrubbed... URL: http://mail.clusterfs.com/pipermail/lustre-discuss/attachments/20060719/10a84582/attachment.html