is there a way to assign spindles in virtual pools?

Scenario

=> Our document imaging process takes documents from electronic fax, email, regular fax, snail mail, etc. from whatever inbound extension (pdf, jpeg, png, etc.) and converts the extensions to single page tiff files.

=> The documents are imported from wherever they come from and sent to a temp location which is attached to a virtual machine in our esx cluster

=> After the documents have been scanned and processed, there is another service or server that will actually read the document from the temp location and begin the conversion process

=> Once the document has been picked up and converted, the document is sent to another location i.e., if the document is read from the volume 12, after conversion the document is sent to volume 8) although logically these are two different locations, the spindles that are being used are the same physical ones.

Inherent problem:

What I am seeing on the back end storage when the conversion process occurs is that the write cache for the VMAX Storage is not being utilized at all. However, I am seeing a huge amount of reads and writes to the disks the scanned into. 

To summarize, although there appears to be a difference between the locations on the servers, there is no difference in the spindles that are used on the back end storage.

=> Solution

Find a way to assign data to different disk groups and or sets of disk spindles in the vmax

This is going to be difficult in the way that with virtual pools, there is no distinct way to assign a lun to a certain set of spindles. This process would allow the conversion engine to pick up any document from the new temp location which would be a completely different set of spindles from the spindles the converted documents would be written to.  I am very confident that there would be no further decrease in performance but should see an increase in performance

Suggestion #1:

Create a cluster resource on 2602(dr location) and add it to the mcgfile01 cluster.  This would require a replication group to be setup on 2538(production)

This new space would be the temp location where documents are scanned into and or read from

This has the lowest cost but does carry some risk as how the cluster would be configured in regards of failing over

Suggestion #2:

Find more EMC disks and create a fenced lun or new lun that isn’t in a virtual pool on 2538(production location) so that we could specifically assign different spindle to a temp location for documents to scanned to or read from

This process would have a higher cost associated with it as we would need to acquire more disks for both the production and dr locations

Suggestion #3

Add another storage unit to both locations.  We could make another storage device here locally and the same out at the dr site but that would be the highest cost and functionally unsound depending on what type of device we use

Suggestion #4

  Create local disks on vm's

  Add local disks on the vm’s and attach them locally to the vm's.  This will reduce the travel time for the data but will still utilize the same spindles for read and write operations which really doesn't resolve the issue in anyway shape or form

any thoughts?