We see many technology innvovation like thin provisioning, compression ,deduplication.... Would these really needed!! if we manage the data properly!!

We see many storage arrays provide various tier of storage. When an application send a write request and array automatically decide on the selection of the tier storage. The decision of the data protection is statically decided during the creation of the volume. I believe application has more knowledge on the data protection and availability of the data. I believe adding a field to the write or read to give hint on data placement would add more value. A few records of an application wouldn’t be off any importance, so those will be placed on less protected tier-storage. Instead of having a static data protection at LUN level, there should be dynamic protection for a LUN and application should decide on the protection level. This way we could reduce automatic application ware overhead on array side. We see many metadata layers like database, file system, thin provisioning. All these metadata would extra requirement for storage space. Instead we have direct mapping of the storage address to application record details. This means allocation can use MDS(metadata server) for loading the storage details. We could apply the simple de-duplication like adding reference to the existing data ete. The data will be protected based on the application requirement. The capacity is not wasted. The data placement is decided by the application.

This means it is close to Cloud Computing... Any thoughts...