At our lab we deal with a lot of images. Each image is about 4-6Gb average. We want to use Lustre for processing the image and place it on more reliable storage such as 3140/NetApp for backups. I was wondering if anyone has a scheme they are using to easily integrate the 2 technologies together so users don''t know the difference. Our experience with Lustre has been good so far. But we did have crashes where we were not able to access the mount point and the process goes into blockstate. We had to reboot all of our 150 clients. We then had to use the NFS share (which was VERY slow) to serve our processing engines. until Lustre was online Is there a way to make a crash from Lustre seamless -- for instance strategical placement of symbolic link on the NetApp filer and have everyone access the link. The link will be pointing to Lustre and if there is a problem with it change the Link back to NetApp. We tried to setup DRBD but it became very costly because we had to replicate all of our data which is about 300TB. Any solutions, thoughts and/or advise? TIA