Yet another storage pool provisioning/expansion question

Hello experts,

we are about to move to production a new VMware vSphere 4.1 Update 1 production environment using VMware ESXi 4.1 Update 1 hosts and the storage system described below:

==================================================
EMC VNX5100 Dual SP enclosure with Fibre Channel front-end ports running VNX Block OE release 05.31.000.5.011.
15X300GB 15K SAS Disk Drives.
==================================================

We need to migrate the following virtual machines:

- a “large” Windows Server 2008 Enterprise SP2 file server VM configured using a very simple partition layout consisting of a boot partition with OS and additional 700 GB data partition for shared user data (currently about 500 GB). Each partition on the VM corresponds to a separate Virtual Disk on the local VMFS of a VMware ESXi 4.1 Update 1 host

- a dozen of smaller Windows Server 2003/2008 VMs running mixed workloads currently thrown in a couple 600 GB LUNs residing on an “old” IBM DS3400 storage system still performing well but lacking of free usable disk space.

Based on my understanding after taking a look at the “EMC Unified Storage Best Practices for Performance and Availability Common Platform and Block Storage 31.0 - Applied Best Practices” document I understand the following:

==================================================
- EMC recommends the creation and use of storage pools on VNX storage systems because storage pools provide more benefits, flexibility and expandability over standard RAID groups.
- Since our EMC VNX5100 has15X300GB 15K SAS Disk Drives EMC I should provision a new storage pool using five drives (4+1) private RAID 5 Groups building blocks.
- “The most efficient movement of data from cache to the Disk Drives will result from the so called ‘full stripe write’”. That’s why “aligning the majority workload I/O Block Size with the RAID Group stripe size has a general performance benefit”. “Alignment matches the I/O Block Size exactly to the stripe size capacity-wise or as a multiple of the stripe size”. When using five drives (4+1) private RAID 5 Groups, stripe size is 256KB.
==================================================

As part of the process of provisioning new block storage to the EMC VNX5100 system before throwing in the VMs I would like to check with you experts if the below mentioned considerations make sense to you and if out of courtesy you could provide additional observations/recommendations.

Since the total amount of disk space consumed by VMs currently in the usable capacity of a single RAID 5 (4+1) Group I was considering starting with a single storage pool building block to reserve an additional building block made-up of the same number of Disk Drives and RAID type for future expansion.

Now the questions are:

1) When adding a 4+1 building block to the storage pool, will the system automatically re-stripe in the background the LUNs created prior to the storage pool expansion across all 10 drives or the LUNs will remain striped across the first RG/5 drives only ?
2) If when adding a 4+1 building block to the storage pool the system does not automatically re-stripe in the background the LUNs created prior to the storage pool expansion across all 10 drives so the existing LUNs remain striped across the first RG/5 drives only, based on my understanding I would rather consider provisioning new block storage to the EMC VNX5100 using a couple of 4+1 building blocks in the beginning to stripe all LUNs across all 10 drives instead of 5 drives only.
3) When working with storage pools instead of standard RAID groups, is “the more drives in a RAID group the higher its performance” statement still valid ? If the statement is still valid, based on my understanding that would be another reason to start with a couple of 4+1 building blocks.
4) Can someone explain the “aligning the majority workload I/O Block Size with the RAID Group stripe size has a general performance benefit” statement with regard to workload in a VMware environment ? Is it still relevant in a VMware environment where different VMs running different/mixed workloads exist ?
I would really appreciate if someone could review this discussion and share its considerations/field experience.

Thanks and Regards,

Massimiliano