Part 2 of our series on the people behind Ubiquity goes to Jay Phillips. Jay hails from Grosse Point, Michigan, where he grew up loving bow hunting and martial arts. Jay enjoys making pizza in his own homemade brick oven wood pizza stove and it’s a real treat when he brings in pizza for the whole company.
Jay’s current hobbies include mixed martial arts, carpentry, and metalworking. He created his own gear that he wears to medieval festivals as well as his custom tent and chuck wagon. Jay is an experienced reader in the area of medieval history and writes his own Lord of the Rings fan fiction in his spare time.
On a Friday night, you can find Jay tucked away behind a wall of expended cans of Mountain Dew, playing Dungeons and Dragons or Magic the Gathering with his friends.
Jay works mostly in our networking and support departments, although he’s qualified to handle billing and sales. Anytime you need a hand, just ask for Jay!
At Ubiquity, we specialize in offering geographically diverse datacenter solutions. We currently offer services out of Atlanta, Chicago, Dallas, Los Angeles, New York, and Seattle. In an effort to meet our customer’s insatiable appetite for more locations and more servers, Ubiquity is opening up a Phoenix location.
Starting in Q1 2011 Ubiquity Server Solutions will be offering Phoenix Datacenter space for Phoenix colocation and Phoenix dedicated servers. Located at 3402 E University Drive in Phoenix, AZ, the new facility will feature totally redudant cooling, connectivity, and electrical systems. With 160,000 sqft of room to grow into and a power density of 200 watts per sqft, our Phoenix location will offer enough power density to keep prices low and allow us to maximize the amount of customers we can fit in the building.
The building is manned by security personnel 24/7/365. Datacenter floor access is controlled by a biometric scanner, and the entire building is under camera surveillance. Multi-factor authentication systems also serve to keep your data and equipment space.
Keep Ubiquity in mind for all of your Phoenix colocation and Phoenix dedicated server needs.
Last time we talked about some things to consider when shopping for hardware. Here are two more.
HDD – The vast majority of servers today use 7200rpm SATAII HDDs. There’s no real good reason to use anything less (like an IDE drive) because the price of SATAII is so competitive with the few legacy IDE drives that you can still find. The only reason to use IDE is if you’re stuck with it because your server can’t support it, and even then, it’s worth considering adding a SATA controller card just for the upgrade. Older drives often have a lot of power-on hours, which helps grind down the lifespan. Older drives tend to fail at higher rates than drives that are only a few months old. Caveat: really new drives tend to fail at high rates, too. The failure rate of drives is actually what is called a “bathtub curve.” When drives come off the line, there is a high failure rate at first, then a long, steady period with a low number of failures, and a few years out, another rate of high failure. Here’s a great article on the topic with a simple graph showing what I’m talking about. So with HDDs, keep in mind new isn’t always better, but don’t get drives that are too old.
Power – The electrical draw of your server probably only matters if you’re shopping for colocation. As a general rule of thumb, with two servers with roughly equal quantities of innards (same number of sticks of memory, same number of drives, same number of processors, etc) , the older system will use more power. Hardware vendors have spent a lot of effort and research building greener machines. Datacenters may be a ridiculously un-green industry, but to our defense, the power we’re now packing per watt of energy consumption would make turn of the century datacenter techs jealous. If power consumption is a concern, you’ll probably actually want to go newer!
If you have any questions, just let us know!
Maybe you’re thinking about picking up some old hardware for colocation. Maybe you’re looking at dedicated servers and you want a good comparison of the server models of yesteryear with this year’s server models. What’s the difference between them? Well, needless to say, new is almost always better, but let’s break down the decision points.
Processor – This is about as cut and dry as it gets. The processor, just because it’s newer, doesn’t mean it’s more powerful. “Power” in a processor is a little bit of a confusing topic for a lot of people, so let’s break it down. Think about what you’re going to be using the server for. Can it use multiple cores? If you don’t know, contact your software vendor for more information. If it can’t use multiple cores, it might not be useful for you to have a shiny new quadcore processor that is 2.2ghz if you can have an older processor that is 2.8ghz. If you can use multiple cores, it’s almost certainly worth the upgrade. Check out the benchmarks of both processors and weigh in things like the number of cores and the clockspeed and how your software is handled by the processor. Most people were taught many years ago to look at consumer buzzwords like “gigahertz” that don’t necessarily reflect the performance you’ll get from the machine.
Memory – This decision goes in tandem with the processor decision. Typically, when you select the processor or the memory, you really limit your options for the other one. For example, you can’t decide to go with DDR3 and a processor made in 2006. The main thing to keep in mind is that the newer the server is, generally speaking, the faster the memory is. The oldest servers you’ll see around in widespread use have DDR memory. This stands for “double data rate;” the memory is twice as fast as it’s predecessor, SDRAM (single data rate.) Most servers in use today use DDR2 memory, which is twice as fast as it’s predecessor, DDR. The newest servers on the market today use DDR3, which is twice as fast as DDR2, and thus 4x as fast as DDR. So when you think about it, upgrading from that server you got in 2006 to a server that just came out this year means you can speed up your memory read/write speeds by 400%! So even if you’re only using 1 core of that new quad core processor, it might be worth it just for the snappier RAM.
Next time we’ll talk about HDDs and power.
So you’re wondering with 1 mbps 95th percentile could possible mean? It means you can use the full 100mbps connection, but 95% of the time you’re at or below 1mbps of actual usage. You’ll hear varying guesses of how much data you can actually transfer, but I usually estimate that each mbps is 330GB up or down.
Think of it in terms of water.
Your provider is offering to give you a 100 gallons per second pipe to your house. You’re welcome to use that entire pipe should you have the need for a lot of water at any given time, which is handy because that allows you to shower, water your lawn, and run the dish washer all at the same time. If you only had a 1 gallon per second pipe to your house, you wouldn’t be able to multitask like that.
But lets face it, most of the time (say…23 hours a day) you’re not using 100 gallons per second, so you don’t want to be billed for that much. There are two ways the water company can measure your water usage. They can either count the gallons you actually use (like metered transit plans measured in GB transferred) or they can bill based on the the quantity you’re usually using at any given time, such as 5 gallons per second (like 95th percentile measuring bandwidth in mbps.)
The reason datacenters might want to measure in megabits per second is because it’s the best way to do capacity planning.
