I think what Jeff has brought up about the possible midi latency that happens when you ask too much of that technology. I'm old enough to have dealt with using complicated midi systems , multiple modules and what not. 
I would never go back to using external midi. Timing is super important and while some people may not notice those little delays, I do.
To me every sound that is available on an external module now has a better replacement as a VST, somewhere. Sure finding the right ones is a big task but well worth the time spent.  
 
My old Korg 05RW sounds amazing, but I would never think to use it anymore and it sits and gathers dust. It cost me $1,500 so I just can bare to toss it out. My Grandkids will find it 20 years from now and wonder what it's for. 

To route the MIDI in Sonar you need to treat each of the synth's MIDI in ports as if it's a separate instrument.

Which means picking the relevant synth MIDI in port as a MIDI track's output port and then setting the track's outgoing MIDI channel to whichever channel the sound you want is on. To record MIDI or play the synth, set the MIDI track's input port to the Alesis controller. You'll need to watch out which tracks are armed for recording and what's happening with MIDi track echoing to avoid MIDI being sent to the wrong place in the same way those of use using several hardware synths do.

As an example -

I have a DSI Mopho, Waldorf microQ and a Waldorf Pulse 2 all using 5-pin MIDI cables which I control off the same USB keyboard controller. The MIDI cables are connected to the computer by a couple of MIDI-USB interfaces each with several ports.

To do this I set the Mopho to channel 1, the Q to channels 2-5 and the Pulse to 6. It doesn't matter what channel the controller's set to.

I then create MIDI tracks as follows-
1. Mopho
2 Q
3 Q
4 Q
5 Q
6 Pulse

Each track output points at the MIDI interface port the relevant synth's MIDI cables are attached to. The input of each track points to the controller.

To play/record I select (highlight) the relevant MIDI track and turn on the track echo button. MIDI will then be routed by Sonar from the controller to the synth via the MIDI track. To play a different instrument or Q voice I simply switch track echo off on the track I was using and activate it on the track I now want to send MIDI through.

The main thing is not to try to record more than one of those MIDI tracks at a time, and to make sure only one at a time has input echo enabled (unless I want two synths to play the same notes of course).

If I wanted to I could even set the Pulse, Mopho and one Q voice to all use channel 1 and the routing would still work out OK. Which is very useful as I have an MD-20mini and a couple of other more oddball things that are fixed on MIDI channel 1 and can't be changed.

Edited to add - if, having recorded the MIDI from the controller I then want to add in CCs generated by the controls on each synth all I have to do is switch the MIDI track input over to the port attached to the synth's MIDI output and overdub the knob turning.

tlw's post here is not so relevant here though. The OP wants 32 channels over 2 midi ports not 5 or 6 over one.
 
Good point too about the Alesis providing one of the midi outs already. I forgot about that. All you need is a second simple USB to midi interface which can be had for very little money.
 
Cactus is not quite right either.  It depends on the quality of the hardware. I have got 9 external devices and most of them are very serious sounding. 2 of them Kurzweils. Many virtual instruments are still struggling to sound half as good. Others I have too like the Kawai K5000W I have not heard any virtual instrument yet that even comes close! The Roland JD800 is a pretty tall order to beat as well. Fat is an understatement. The fact that a machine like this has a huge front panel allowing you to create and sculpt sounds is also a revelation. Something virtual instruments are just not really capable of. Even a Roland JV2080 has a rather unique sound to it as well. Not easily emulated either.
 
But I also admit that the fantastic thing about virtual instruments is they have the ability to head into territory no hardware device can even think about. e.g. Iris 2, Prism, Granite or Mangle etc...I do find that exciting. I find the blend of hardware and software rather interesting and incredible.  A sound you won't hear with just software alone.
The advantage of having a serious hardware setup is that a wall of sound can be had before any virtual instrument is even loaded into your DAW. Meaning your DAW is like walking in the park CPU wise.
 
Midi timing is better in Studio One as well. Especially if you use all 8 midi ports and if you have them only sequencing 1 or 2 parts down each port. Actually I am only addressing one instrument per port most of the time which is luxury. (1ms latency which many virtual instruments are struggling to achieve) I do feel the response of the hardware is almost second to none too. Except these days I also have got a Mac and I am running Studio One on that and this time with a thunderbolt interface with 1.2 mS latency. Only now I am feeling this is as fast as the hardware.
 
 
 

Wow - Thank you everyone so much for the responses and guidance , it is really appreciated - 
 
From what I can gather, I have 3 options
1. Get another midi controller Keyboard and use each controller keyboard midi out to the Xv midi in1 and midi in2 and usb from keyboards to pc
2. Buy a USB to midi interface cable- any recommendations I see they go from $29 AUS upwards
3. Buy a Hardware midi interface - any recommendations
 
thanks again all 
 

This is better served in Hardware...

Jeff Evans
Midi timing is better in Studio One as well. Especially if you use all 8 midi ports and if you have them only sequencing 1 or 2 parts down each port. Actually I am only addressing one instrument per port most of the time which is luxury. (1ms latency which many virtual instruments are struggling to achieve) I do feel the response of the hardware is almost second to none too. Except these days I also have got a Mac and I am running Studio One on that and this time with a thunderbolt interface with 1.2 mS latency. Only now I am feeling this is as fast as the hardware.

When using MIDI cables...
1 note ("On" or "Off") transfer takes 1ms (~30 kBit / 10bit (MIDI byte + start/stop) = 3000 MIDI Bytes/Second / 3 Bytes per note On/Off = 1000 Notes per Second = 1ms
So, 10 fingers chord takes 10ms to transfer.
16 instruments throw one cable have a jitter 16ms when they want to play one (!) note each at the same time.
 
Hardware synth do not have zero latency, more like several mS.
 
Most synth sounds do not have attack under 1ms, the difference between instruments can be huge.
 
What I mean, +3-5ms latency for anything except "tactile" direct feedback with rapid attack (voice, guitar, drums) is physically impossible to distinguish.
 
Do not get me wrong, I also prefer to play my DP with local sound and clearly notice the difference when switching to DP->USB->PC->Soft synth->Slow audio interface (10ms+). But what I claim is that changing the buffer and so latency in the last component (audio interface) in range under 5-8ms is only possible to notice when local sound is played in parallel.
I mean (DP->USB->PC->Soft synth) or even worse (DP->MIDI->(USB/thunderbolt)->PC->Soft synth) latency is far from 1ms (some USB MIDI keyboards, like Roland, try to reduce at least the first part with "hi speed" MIDI mode/drivers, but most do not bother).
 
 

The main point about my post was do not right off the sound of the hardware.  Saying that virtual instruments sound so superior is just plain wrong.  Remember all the virtual instruments out there started trying to emulate all this hardware in the first place. 
 
Cactus should fire up his Korg synth.  He may be surprised exactly how great it does sound.  It is easy to forget. In fact if I had one of those I would be using it.
 
I am someone who uses both hardware and software side by side all the time so I get to hear both and how both can sound very good.  It still amazes me how great the Kurzweil or my EMU sampler actually sounds.  The detail, depth, power, fatness.  Or the JV2080 or the other synths I have. The best the software can do when trying to emulate hardware is basically match it.  And in some cases it does.  Where the software really excels for me though is its ability to create sounds that no hardware synth can even go.  That is very cool indeed. 
 
To azlow3, you are not using a Mac.  I can because my software runs on both platforms.  Thunderbolt is fast and I mean fast!  I can monitor direct through the software for example a vocal track without any perceivable delay of any description.  I can feel the response is just that little bit better with virtual instruments over thunderbolt.  Especially percussion sounds.  I am very sensitive to this as I play drums as well as keys.  When you hit a real drum with a stick there is NO latency of any description.  It is the ultimate response experience. 
 
The OP has got a great instrument as well.  There is not much the XV5080 cannot do really.  (Fully expanded!) Anyway I think we have solved his problem of getting two MIDI ports feeding the two Midi INS on his XV synth.  That is the main issue.  
 
Sonar can drive any number of Midi ports as well.  Previously I had 8.5 Producer running with an Emagic Unitor 8 Midi interface (8 Midi Ports) and it all worked flawlessly. So I was addressing 128 Midi channels at the time.