Chiming onto this thread late, I've definitely seen customer workloads which used a whole fleet-load of blades connected to a SAN network, the idea being that the blades were easily replaceable, and could take various different responsibilities due to shifting workload, battle damage, etc. So this would be a case where the individual blades wouldn't have huge amounts of memory, but there would be a huge number of block devices available on the SAN network. So at least in this case, avoiding creating every single BDI thread until the block devices is first used would be a major win for this sort of customer deployment. -- Ted -- To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
