Doubt it would lower the cost too much. I don't really find the license cost per cpu to be a big factor since they charge per actual physical CPU. If you buy multi core CPU's like Intel E5-2620's to build your grid with the cost per CPU is pretty decent.
OK, i'm thinking that my question was maybe badly worded...
Lets say "AppLogic 3.7's CPU oversubscription feature has brought the licensing & hardware costs within range for smaller VPS hosts"
For example Techark, using your own "Cloud Business" package specification of:
1 CPU core
1 G RAM
30 GB disk space
Assuming you're using dual 6-core, E5-2620 CPUs with hyper threading on, then you'd be able to support 24 customers (more or less) on the following server specification:
2 x 6-core CPU
2 x 1TB 10K SATA HDD
Lets say, ballpark figure, that server spec is going to cost around ~ $4,500 (keeping the numbers simple), plus (list price) AppLogic licenses of 2 x $80 per month (based on annual subscription licensing model).
So per customer (without CPU over-subscription):
Annual AppLogic license cost per customer = ~ $80
Hardware purchase costs per customer = ~ $187
So with CPU overs subscription of x2, you'd (theoretically) be able to support 48 customers, with the same VPS spec, on the following server spec:
2 x 6-core CPU
3 x 1TB 10K SATA HDD
Ballpark purchase figure for hardware, around $5,000
So per customer (wit CPU over-subscription):
Annual AppLogic license cost per customer = ~ $40
Hardware purchase costs per customer = ~ $104
OK the numbers can be + a bit here and - a bit there but hopefully my point is clearer..
I see what you are aiming at but no you are not going to get there at the price point you are wanting using Applogic.
I suggest you join the Applogic forums or the Applogic Linkden group which is quite active. I think you will find a lot of your answers there.
Your numbers are still off by a quite a bit because you are not taking into account the spare capacity that you have to have as part of the grid. Plus the network and power redundancy you would want for a cloud grid.
In the OP's scenario, the overhead involved in network and power redundancy along with spare capacity really aren't variables that factor in to his question - the cost for those things is the same, regardless of the CPU over-subscription. Network redundancy, power redundancy, and spare capacity would need to be considerations regardless of whether you were sticking 24 VM's per node or 48 VM's per node.
Unfortunately I don't have the answer to the OP's question since I'm not an Applogic user, so hopefully someone can address that specifically. I just didn't feel it was right to dismiss the inquiry out of hand for inconsequential reasons.
Rob Tyree Versaweb - DDoS Protected Cloud and Dedicated Server Hosting Fiberhub - SAS70 Type-II Colocation in Las Vegas and Seattle