As you'll likely see on the site, we're making a huge shift here at the company. There is a big problem with our planet's environment, and until recently, SingleHop hasn't even taken notice to the additional harm that we're doing! We don't claim to have the answer, and we don't claim to be environmentally friendly, but after taking note of the thousands of tons of CO2 we're producing, we've already began making changes, and we're going to continue to do so!
As many of you know, for approximately a year we have been on a data center hunt, which originally began in Chicago, and I quickly found myself, due to lack of space, looking in other parts of the country. I actually shouldn't say space, because if there is one thing that we have plenty of in Chicago, it's space. The problem was available power. Server manufacturers keep putting out hardware components that just drain power like there's no tomorrow. The reason is obvious- it keeps their costs low, and then when companies like ours buy their components, it keeps our CapEx low as well, which in turn keeps your costs low. What an incredibly blind move, because according to studies it costs on average (this was determined by the lifetime of the hardware) 2500% of the CapEx associated with purchasing hardware to actually power it. Taking into account that we have millions of dollars in hardware, and our hardware expenditures go up every year- we're talking a lot of money here in power, and the hardware costs become less of an issue.
Back to my story... On our data center hunt we kept running into the same problem. Our dedicated servers were pushing too much power for most data centers to accommodate. After months of this, and researching online, I found this wasn't a problem that was limited to our company. In 2006, 60% of data centers ran out of power, and when they did, they didn't expect it at all. While I'm not an electrician, I think that has to do with power consumptions on servers increasing every year, and growth of the Internet and data center demand. At the time when we were buying servers we honestly didn't put any thought into the amount of power we were consuming. Power was cheap enough to where we never factored it in, and because of that, there wasn't a demand for power efficient components, so they weren't being produced.
Now that power is a huge problem, technology vendors have taken notice, and are stepping up their game, and so are we, to the extent that technology allows us to. After finally finding our data center, we went into testing to determine the most power-efficient servers. Right now there are a number of different vendors that have developed more efficient products, but there isn't anything that off the shelf that brings these components together, and gives you what we felt would be as power efficient as possible. After months of testing various components, we were left with a server line that in comparison to our old power hogs would cost an additional ~20% in CapEx, but the results are well worth it. We've decreased the power draw on dual processor machines by as much as 1.5 amps, and single processor machines by nearly an amp. For those that don't know, a single amp of power is approximately $10-$12 a month in power costs, so we're talking some serious savings here over time. Environmental issues aside, it also means a power infrastructure with less strain on it, and more capacity to grow.
While our data center, and power struggle was a difficult thing to go through, I think that we've come out a better company that is positioned well for the long haul. Going forward, you can expect to see us deploying servers that have the highest speed-to-power consumption ratio (the most efficient!), because let's face it- speed is very important to all of us, and as we become more educated, we'll do our best in keeping you informed.