Want to plan an upgrade from Celerra NX4 to VNX for early 2014, some questions.

I'm gathering information to research, plan, and get budgetary info for an early 2014 approval for our infrastructure improvements.

Currently we run VMWare ESX 4.1 U3 across 3 Dell R710 hosts.  They are running about 45 active VM's and each host is about 80% capacity in their 64GB Memory area.  There are 4 NFS datastores housed on a Celerra NX4 using SAS drives.  The communication between the VMWare hosts and NX4 is on it's own VLAN and connected via 1gbps Cisco 3750g switch.

At a remote site we have another NX4 but with SATA drives.  We run Celerra replicator from our main HQ to the remote site.  There is one VMWare host here running VMWare SRM and again a Cisco 3750g switch with a similar configuration.

I'm looking to ADD 2 new VM hosts at HQ in addition to add a VNX to migrate to.  After migration to the VNX, we are debating keeping the NX4 and using it to keep backups, while using the new VNX for Production.  Not sure if we will sell it back to EMC (though they did offer us a credit if we do).  Here's my questions to ya'll...

Can the VNX replicate to our existing NX4 at our DR site?  Would the replication have to start fresh?  It's a 50mbps fiber link between HQ and DR site but we have a schedule in place so Celerra replicator uses 20mbps during business hours, then is unlimited off business hours.would

Can we get a new VNX with 10gbe interfaces for both data movers?  What would be a good switch infrastructure?  A Cisco Nexus 2000 series?  Were familiar with Cisco as all our networking gear from core to wan to wireless is Cisco.  What type of cabling (copper, fiber, etc..)?

I'm thinking of adding a 10GbE switch, and adding 10GbE PCI cards in the current hosts and ordering new hosts with 10GbE cards as well.  The new switch and all 10GbE gear would be entirely separate and jumbo frames turned on.  We currently do not use Jumbo frames and I don't plan on altering that portion of the network because I do not know what that would do if not everything supported it (from clients to routers to wireless to other ancillary switches).  Doesn't seem like you can turn on Jumbo frames on a per VLAN basis... its all or nothing.

So I'm looking at an entirely separate 10GbE backend for storage between all VM Hosts and the new VNX.  Then keeping existing Cisco 3750g for VMWare hosts connectivity to our core which would be various vlans for voice, data, dmz, vmotion, etc.  The current hosts have 4 onboard 1gig ports and a card with 4 1gig ports (not all used today).  I'm just thinking to keep the new VNX completely physically separate  (Except for a management interface and replication interface).  I also plan on upgrading the VMWare 4.1 to 5.1 or 5.5 if it's available at that time.

Any great recommendations on connectivity, if the VNX can replicate to NX4's, etc... that would be great.

Reasons for shopping:

1.  Celerra NX4 support more than doubled because its now considered older gear.  Nearly $12,000 a year for NBD support vs $5000 previously.

2.  Starting to hit performance ceiling.  Seeing some datastores spike from 40 to 300 ms latency from vm host to nfs datastore.  Though normal average is under 16 ms... these spikes depend on whats going on in the environment and what datastore were looking at.  Interested in the VNX flash cache also paired with possibly a new 10gbe connectivity to vmware hosts.

3.  Current vm hosts are at 80% capacity.  We also would like to continue to grow and while adding VMWare hosts would be the next step, we also want to make sure the backend storage is ready for the added connections.  This also has me thinking of upgrading the storage connectivity to 10Gbe rather than 1Gb.  We realize that VMWare's NFS implementation can't aggregate bandwidth on NFS connections across LACP links unless we do some creative multi-subnetting, multi-interface, multi-vlan configuration.  Going to a straight 10GbE interface across 2 stand alone switches (like the Nexus line) for failover might be a better solution with less complexity.

THANKS!