It seems that IBM DS 3512 supports synchronous mirroring. However in order to acheive what you need the two nodes must be smart enough to detect split-brain conditions, which i couldn't find any documentation/indications for. Let me explain if the synchronous mirroring were to support split-brain and assuming that a single ESXi host can acces both the LUNS then
ESXi writes to node A
node A mirrors data to node B LUN and writes to its own LUN
Sends the response back to ESXi
If ESXi detects that node A has failed then ESXi switches over to node B.
This is easily done by the remote licensing you have. The issues creep in when
ESXi writes to node A
node A cannot communicate with node B, thinks its dead and writes to only its LUN
now node A fails
ESXi switches over to node B. data is already corrupt
The above is just one scenario. There are a whole lot of complex issues than above. Others are correct, you need additional software to achieve what you need. However you can check with IBM support if they support active-active IO. If they do then they are handling split-brain scenarios. I rather doubt it, though.