Recently I read an article that discussed carbon emissions and our efforts to reduce them here in the US. The article pointed out that in China alone, runaway fires in coal mines burn as much as 200 million tons of coal each year. That would be nearly 10% of the coal China actually uses. Now we know that burning fossil fuel emit carbon dioxide (CO2) and burning this amount of coal would emit 1,120 million metric tons it! This is the same amount of CO2 emitted by gasoline burned here in the US each year. The difference of course is that we get productive work from most of our burning of gasoline. The uncontrolled burning of coal while still in the mine doesn't do any good for anyone. In fact it prevents mining of that coal and nearby coal as well (and that's in addition to the CO2 emitted.)  
 
Here in the US many legislative bodies are trying to come up with ways to reduce our CO2 emissions. Some of these ideas are to increase our average fuel economy and coming up with alternative 'carbon neutral' fuels. But these measures will be slow to act and are also being fought by others such as the Detroit-3 (GM, Ford and the Chrysler group.)  
 
Whether or not you believe these emissions are causing global warming or not it's pretty obvious that if we simply went over to China and helped them put out those fires the potential CO2 reductions are absolutely huge! Look at all the good will we could get from it too. OK Sure maybe we should do both, reduce our emissions AND help the Chinese put out the fires. But I'm not trying to put a Carbon Guilt Trip on anyone, just pointing out that there are different ways to look at some things. Here's another simpler example. I used to drive a 4x4 Dodge (Old Red) as my daily driver. It got maybe 15 MPG and as gas prices went up all my driving added up to a lot of money! At $3.50 a gallon I could sink nearly a hundred bucks into the tank every few days. So I drove more carefully and was able to get the mileage up to about 18! Woo Woo! I could stretch the tank another 60 or more miles, certainly an improvement. But Old Red was getting tired so instead of another big vehicle I chose a Toyota Prius. At more than 50 MPG I can go nearly twice as far on about half the money. It wasn't maximizing the MPG of the Dodge that saved me money it was replacing it entirely. And I got a better ride, nicer stereo, and some way cool computer displays in the deal!  
 
There are different ways to look at such things like the cost of your Computing systems as well. When you decide on which servers to put into your data center which numbers do you look at? In a nod to CO2 emissions, or maybe your power budget, do you look at how much power will be consumed by your infrastructure? This can mean more than you think as more power means more or larger UPSs, a larger generator, and of course a bigger power bill. And here's a clue too: do you know the simplest way to know how much Air Conditioning that your server room requires? Just check the output power being supplied by your UPS. This is because virtually 100 percent of the electricity that goes into your server room ends up as heat. More power, more heat, bigger air conditioning units, and of course bigger A/C Units draw more power as well further increasing the size of your power bill (and your generator!) If you need a dozen windows servers, KVM connections for them all, enough outlets to plug them all in, switch ports for Ethernet and backup drives and of course disk storage for all of them instead of a single System i will you have enough power and cooling? As important as power is, is this the trip wire number when deciding on which server? Probably not.  
 
Even with today's shrunken servers your choice of server may affect the physical space requirements in your system room. Will you have enough space for that dozen windows servers and all the stuff listed above? What about a single System i? Another important decision to be sure but it is THE question to ask? Again, I would say no. What about acquisition cost. If you purchase those 12 windows servers and supporting accoutrements will that be cheaper than the one System i? Maybe so. Suppose that they are in fact cheaper to acquire. Is that truly the proper metric to use in the decision of which server solution to choose? Here comes that familiar answer again, NO. What about the utility software required to run your solution. With windows you need a good solid (really!) anti-virus solution. You'll need good backup software. You'll may want software to help you roll out fixes or close the servers for each rebuild. You may also need additional security software. i5/OS doesn't get viruses and it's backup and security systems are built in. Most important cost savings? Not yet.  
 
It's maintenance right? You gotta account for maintenance of the servers. So what is it going to cost for server maintenance for those 12 servers? I know you'll just use the warranty and then repair them on your own while the System i is going to require a contract. Still, this isn't the most important criteria either.  
 
So what is the right metric then, haven't we looked at all the important ones? Are you ready? NO! There are two general types of expenses, one time charges and recurring charges. It's almost always the recurring charges that cost you the most money. So what is the most important recurring charge? Are you ready? It's People! For the cost of one system administrator for one year you could purchase FIVE model 515s! And while those 515s run for three years the administrator will cost you enough for TEN more! So who wins the battle of staff? Like the coal mine fires vs. gasoline consumption in the CO2 race, windows servers require hugely more staff than System i. If you have 12 servers vs. just one, you can imagine that every administrative task that needs to be done will be done 12 times as often, maybe more depending on the task. And then there are the tasks that just don't need to be done at all on System i. Recently we had two customers who ran the same software one on System i the other on Windows. The question was how many DBAs do you each have: The System i customer has none, the Windows and Oracle customer has six. Take the cost of six Oracle DBAs out of your staff and you can do some good things in the server room!  
 
Over and over and over we hear about AS/400s and iSeries and System i servers that are run by one part time person while the windows side of the staff just grows and grows. Then some one looking at the wrong numbers declares the System i is too expensive.  
 
Take a look at your staff and see what they do. How many staff dollars actually go into supporting each of your systems? Add that in and look at the real costs. Is Linux really free, is windows cheaper, or is System i the real winner? Are you getting productive work from your IT Dollar or are you just burning money in the computer room?  
 
Are you looking at the right numbers?  
 
- DrF