What is the difference between a data center and a location to meet network infrastructure needs?
To make the best decision about how to meet your business network infrastructure needs, you must understand the difference between a data center and a co-location arrangement. What can each do for you … and what can’t you do?
In some respects, they are all the same. Colocation simply means co-locating your network equipment in another location. CLEC teams (FD, XO, Megapath) place the team all the time in an ILEC CO or POP (Bellsouth, Verizon, Quest). They are simply placing different teams, but the end result is the same.
In reality, at one point, MediaOne in Atlanta (long before they were taken over by AT&T and then Comcast) allowed anyone to place equipment at certain central sites. I don’t know the details of how, what, where and why, but I know it was done. Theoretically, the same could be done at Bellsouth or similar if you know the right person or are willing to pay the right amount, which may have been the case with MediaOne.
For Data Center vs Collo, they are mostly the same. Let’s put it this way, a colloquial facility is a data center, but the reverse may or may not be true. The only difference between the two would come into play in the data center rules. Some data centers (e.g. NAC, the site where DSLR is hosted) may not allow equipment placement (I don’t know if they do or not, it’s just an example) but they will allow you to buy all your pre-existing dedicated equipment. servers you want and they basically accomplish the same thing. But assuming that NAC allows you to place equipment doesn’t mean it’s not a data center.
Basically, a data center is any hardened facility (we hope) that houses various types of equipment in order to allow remote users to access it for various reasons or methods. A CO or POP could even be called data centers, in fact they are likely to be much more hardened than a typical data center. But just as a CO or POP can meet specialized needs, so can a setter. Basically what I mean by that is that a CO would generally not host your server, but it will host your DSLAM if it were a CLEC, a colo on the same token will host your server, but you may not have pre-existing dedicated servers to sell. you.
As for how you get the bandwidth you need … that really depends on where you put your servers and that also means you may or may not be billed at the 95th percentile as well. If you buy rack space at a transportation hotel like 55/56 Marietta here in Atlanta, it would be in the 95th percentile because it is connected to the building’s network infrastructure. The upside to that is that there are dozens of companies that have endpoints in that building, so say you wanted something from Georgia Tech servers, just a couple hops on the colo facilities and it’s on the gatech.edu network and it hasn’t even touched the public internet to get there.
But at the same time, if you rent space from an ISP, then you have the advantages of power and cooling, but you are not directly connected to your network. Instead what they did was sell you the T1 or DS3 or whatever and all you paid was the cost of the port with no loop costs because there really was no loop (at least not in the sense of what is on the poles ).
There is also a difference in how each company sells its space. Some places allow you to buy 1U of rack space at a time. Some allow you to host tower-based servers, while others do not. Some force you to buy a portion (or all) of a shelf (say 1/4 or 1/2 of a shelf) but it will be dedicated to whatever you can fit in there and will generally be locked. XO, on the other hand, only sold cages. What you get is basically a little “room” where you bring your own shelves and fill them however you want. This is generally the best option for those who have existing racks and infrastructure and just want to have it in a data center.
Remember … placement is not about staffing, it is about the location of your equipment. Rather than running data circuits and upgraded power supplies in your location, you will place your servers in a data center where they have plenty of high-speed circuits and robust environmental controls, and large backup power generators. They have staff in those data centers, who will sometimes do things for you if hired at an additional cost.
The other side of Collo is bandwidth. If you say you are a “bigger” company … the construction cost of running a DS3 or more at your location could be prohibitive. This is more directly the point if you need peering redundancies.
The biggest difference is if you “rent” bandwidth instead of your own pipeline. Most bill at 95%. It’s not like an all-you-can-eat DSL or cable line. You pay per MB … plus the rack space you use and power. The silver lining is that the world could end, but your servers are most likely fine. COLOs are built like Fort Knox. However, if the world is over, no one will care that their servers are up.
If you need help finding a colocation solution … or any bandwidth solution for your network infrastructure needs … you can get that help free of charge at DS3-Bandwidth.com.
