| How would you respond to someone telling you that | | | | mounted servers, running 24 hours a day, 365 days a |
| you could reduce your operational costs by 50% or | | | | year. The standard air conditioning needs would be |
| more? Would you believe them? Would you throw | | | | provided by an 8-Ton unit, capable of handling the |
| them out of your office? In this series of articles we | | | | output of all 10 servers. The annual electrical cost to |
| will examine just that. | | | | run each server would be approximately $975 per |
| We've all heard the mantra of going "Green" and how | | | | year. In addition, the cost to the company to cool |
| that will save money, but do you know how you can | | | | the heat put out by each server would be another |
| go Green? What's involved, and what can you | | | | $410 per year. This makes the annual cost of running |
| reasonably expect in the form of savings? Can you | | | | each server a hefty $1,385. Now multiple that |
| implement a program that not only uses energy | | | | amount by 10 (the num of servers in the |
| resources more effectively, but really saves your | | | | environment), and we see that it costs |
| corporation some of its own "green" (dollars that is). | | | | approximately $13,850 every year to run your |
| These are the questions that IT professionals and | | | | datacenter. |
| business operations people ask every day, and the | | | | Solution 1 - Upgrade your hardware |
| answer is definitely yes. | | | | The first solution is to simply upgrade your hardware. |
| Here's a great beginning to going Green, and an idea | | | | Many manufacturers have gone on the Green |
| of which methods gives you the biggest bang for | | | | bandwagon and have done a great deal to reduce |
| the buck. Here's another surprise: You do NOT need | | | | the electrical consumption of their products. Savings |
| to have a hundred servers in order to reduce your | | | | can be achieved on both the amount of electricity |
| Total Cost of Ownership (TCO) and save substantial | | | | consumed and the amount heat being produced, |
| amounts of money with your Green initiative. Even | | | | simply by changing the processor and the drive types |
| companies in the Small to Medium Business (SMB) size | | | | selection. |
| can demonstrate substantial savings. Our example will | | | | For example, by choosing a Rack server with SAS |
| be scaled down to the SMB environment to | | | | drives instead of SCSI drives, will save you about |
| demonstrate what can be saved on the bottom line | | | | $121 a year. Now multiply that by the number of |
| annually. | | | | servers in the datacenter, and you can save over |
| Simple MathIn an effort to keep the math from | | | | $1,200 a year just by selecting lower consumption |
| clouding the results, we'll speak in generalizations. A | | | | drives. |
| spreadsheet to be used to calculate the results, for | | | | Believe it or not, one of the biggest consumers of |
| those that want to see how we structured the | | | | electricity and producers of heat is the CPU. If we |
| savings, can be obtained by contacting the author. . | | | | use our standard Rack server (with two physical |
| If you would like to run the equation against your | | | | CPUs in the server) and use 2 Quad-Core 3.5GHz |
| company's actual consumption; definitely reach out to | | | | CPUs (instead of the single core 3.7 GHz CPU), we |
| the author as well, to be assisted with those details. | | | | realize a savings of approximately $460 per year. |
| We'll begin the equation with a baseline of what a | | | | You will save an amazing 30% off your cooling and |
| traditional IT infrastructure looks like. What we often | | | | electrical consumption simply by using the modern |
| find in corporations is a host of specialty applications | | | | quad core processor design. |
| running inefficiently alongside one another: Customer | | | | This is a win-win, since you are reducing the |
| Relations packages, Email, and many other | | | | operational cost of the server and your company |
| server-based applications. Basically, each vendor | | | | also benefits from the increased performance of the |
| requests a dedicated server for one or more if its | | | | quad-processor design. And the exponential impact |
| components. This quickly leads to a plethora of | | | | this makes on our national energy consumption is |
| servers being added to your environment, ALL | | | | huge. At the end of the day this decision is a |
| running their single, specialty application. | | | | no-brainer. |
| Should you be daring and sidestep their | | | | Conclusions |
| recommendation, adding their application on a server | | | | In this section of our series, "Going Green in the |
| with another application, you open yourself up to the | | | | Datacenter," we have seen how the drive type, and |
| ridicule of their "experts" should you need support. | | | | more importantly, the processor selection, can have a |
| ANY service or performance issue will instantly be | | | | direct, long-term impact on your bottom line. In our |
| your fault, based on your decision to install their | | | | 10-server datacenter, we were able to reduce the |
| "perfectly working system" (yeah right!!) on a server | | | | TCO savings by over $4,600 per year simply by |
| with another application. Even if the other system | | | | choosing a more economical CPU. This is equivalent to |
| has nothing to do with your specialty package, trust | | | | saving the cost of a new server every year. |
| me, it will still be your fault when their program does | | | | Sounds easy right? As the TV commercial yells, "But |
| not work correctly. Adding insult to injury, these | | | | wait. There's more!" In the following three parts, we'll |
| servers are massively under utilized. It is not | | | | go much deeper in our Green Datacenter project and |
| uncommon to observe CPU utilizations consistently in | | | | demonstrate that this only scratches the surface. Our |
| the single digits. | | | | goal will be a savings of over 50% in our new Green |
| The "Typical" datacenter: | | | | Datacenter. |
| In our example datacenter, there would be 10 Rack | | | | |