CAT is great. But really there is not enough of it to make it convenient for regular use by more people. We need a real transportation solution that is not only green but incredibly practical. A lot of folks cringe at the thought of using public transportation. But the fact is, it’s a lot less stressful to sit in a subway car or train for an hour or so than it is to navigate I-285 traffic hell for the same amount of time. In fact, having public transportation makes your morning commute a time you can prepare yourself for a good day’s work instead of the time you spend contemplating the consequences of road rage and feeling beat before your day even starts.

If we don’t do something now in Savannah, we are going to be just like Atlanta, that is, if we want to sustain any kind of growth. Savannians are understandably hesitant to fully embrace sprawl-style mass destruction of its aesthetic.  We love our city. We want to grow but we sure as hell don’t want to grow up to be just another nowhere sprawl town (not that we think less of people in those towns. You play the cards you get and if you’re in a sprawl town, I’d suggest you play your cards to find ways to make it better).

We are one of the greenest cities in the U.S. Civic architects come to downtown and to drive under our tree canopies to learn from our example and to contemplate potential for a better way to grow cities. Of course we are far from perfect but the great thing about Savannah is that we are like a history book of civic development — good and bad and everything in between going back almost 300 years.

At this time, we can’t justify something like an elevated monorail system or a mag-lev train (although what do you think a 200 MPH train from here to Atlanta to Jacksonville, Charleston and Charlotte would do for our growth?) and we sure can’t justify a subway system.

BUT WE CAN JUSTIFY PERSONAL RAPID TRANSPORT!!

The cost per mile of track is only about $6 million to $16 million + about $100K to $300k per station. Compare this to the PRESENT cost of building a highway — $19 million to $1 BILLION per mile!

Rather than spend the next 2 pages writing about it, here is a brief video about a PRT for London Heathrow.

A long-running proof of concept came from the West Virginia University, where PRT has been going since the 70’s. Sooo.. keep in mind that the ones in this video are old.

The cost for building a PRT line that runs the length of Abercorn is LESS than expanding it or building some kind of perimeter. Please let us know what you think by posting comments on this article. We will add more links and info to this page later.

kaht Emerging Tech, Green Industry

Computer Power

Before we continue I need to say that this is being written on a laptop, which draws a meager 10 to 15 watts of power. That translates to about 0.012 kilowatts and I would be drawing 0.012 kilowatthours in 1 hour. If the price of 1 kilowatt hour is $0.09, then the cost of running the laptop for an hour is $0.09 * 0.012 = 0.108, or about .11 cents.

If I ran the laptop for about a 10 hour work day then it would pull roughly $0.01 per day. (it’s hard to make a precise calculation because the device’s power demands change based on what you happen to be doing).

What about your honking desktop machine? It pulls a whopping 300 watts of power to do the same thing that your laptop does AND we are more likely to leave it running. How much does it cost per day to run a desktop CPU? $0.09 * 0.3 = $0.0270 per hour, or 3 cents per hour. But typically your newer machines (unless you are running a bot net virus) won’t pull all 300 watts so let’s say about 2 cents per hour.

Over a ten hour day that adds up to 20 cents. If you leave it running 24/7, that is $0.48 per day, which adds up to $14.40 a month.

Which brings us to server power. A server typically pulls more than a desktop because a serious server will have redundant power supplies and more aggressive cooling requirements. But let’s say due to software power management that each server pulls about $15 a month. A room of 10 servers then would pull $150/month, or $1,800 per year. A room of 100 servers, $18,000 per year.

As a foot note, Google runs about 450,000 servers in its facility, which comes out to about $81,000,000 per year. This figure does not take into account the enormous air handling systems and other infrastructure power needs of their data center. In perspective, that’s not a huge price to pay for managing so much knowledge.

Where virtualization fits in

Think of the bees as virtual machines and this box as a blade server. Each blade is a physical server. Each bee is a virtual server.

Chances are, you use your desktop or laptop often just to read email or look at a few web pages. But these machines, about which we frequently complain of their chronic slowness, become slow typically because they are waiting for outside things (like stuff on the net to come down). You are likely using less than 10% of your processor 90% of the time. Right now I have 2 text editors and 4 web browsers running and my CPU utilization barely spikes above 4%.

The wasted CPU cycles and power is scant with a laptop. But where this makes a difference is in a datacenter where there can be thousands of computers along a single wall.

Imagine a 20-story office building with only 20 people working, one on each floor. You still have to run the air handlers and send electricity through each floor of the building to keep those tenants happy. At this point, the building is under-utilized, or below its capacity. Yet the infrastructure still has to run as if the building were full.

What if (and this may never happen with real people) the workers in the building were told  to make the most efficient use of the office spaces? They would all be required to work on one floor when there are only 20 of them there.

However, during the day time when there are 1,000 people working in the building, they will all scatter throughout the floors and pumping power and air through the entire building makes sense. The building is fully utilized, or at its capacity.

As people leave later in the day, people who stay move closer to others who are staying.

The vacant offices and floors can have their power and air handling shut down or at least minimized.

This is how virtualization works to save power. It makes better use of unused space. Simply put: 1 server running at 75% capacity accomplishes the same tasks as 3 servers running t 25% capacity. Yet, all 4 servers (whether running at 25% or 75%) will pull about the same amount of power and require as much cooling to handle generated heat.

Virtualization allows the 3 physical servers working at 25% to move into ONE physical server working at 75%. Yet to the outside world, there is no difference — all 3 servers are still accessible in the same way as they were when they lived in their own physical machines.

That is a 3 to 1 reduction, or consolidation. The power savings are considerable, especially if this scheme is multiplied out over, say 10,000 servers.

Consolidating servers at night and shutting down unused physical machines can save millions in power bills.  That is the immediate effect.

But what if the demands don’t really increase a lot and the server room simply has more physical servers than it needs? Before virtualization for server class PC’s became mainstream, analysts had no choice but to account for future growth by adding more machines in anticipation of the growth. If they are wrong about their expectations, the company just bought a bunch of machines that will not be used, will waste power and will be likely obsolete in a matter of months.

With virtualization, if you had 20 machines in the room and you need to configure 100 separate servers, it can be done without purchasing more physical machines. Often the demand on servers that are built to pre-empt growth will not be used at full capacity from the beginning. Put your beefy existing servers to work at full capacity instead of buying new machines that are just going to run idle and then be obsolete by the time the anticipated growth shows up.

The indirect power savings are enormous. Fewer circuit boards need to be fabricated, fewer cases need to be cast, fewer cables and wires, fewer truck and cargo ship trips to handle  fewer physical servers.

This barely scratches the surface on what virtualization can do for you. See VMWare’s TCO/ROI calculator at Alinean to play with some numbers and contact us if you want to know more.

kaht Green Industry