Expert's answer

My first question would be: With all the different CDN's that are currently in operation, why do you want to design your own? What are they not able to do that designing/operating your own CDN will let you do? These are questions that you will need to be able to answer when going for the financing that you will have to have to purchase all of the equipment (whether you buy or lease) you'll need for this.

Completely answering your questions could fill a book, but I'll try to give you a start for more questions to ask and to continue your research. I would suggest engaging technical people with CDN experience with the various networking vendors, such as Cisco, Foundry, Juniper, etc. If you are going to do this on your own, you need to talk to the various vendors that are out there to see who has the best options for your specific situation.

One of the things you will need to look closely at the the different types of video encoding that are available. Distributing 30-minute video clips, for example, will require considerable thought. The more you can compress the video clip, the faster it will download. Something else to consider is to having the clips downloaded during off peak hours when you may have an easier time at downloading a large file.

One concern I have is that you may have less than stellar performance out of a DSL connection at the customer end. Even if they order a "business" grade DSL connection, they won't get the same level of response/support as they would expect to get with a T1 connection. One of the biggest challenges you will have with a DSL connection is that it won't have the same upload and download speeds unless you order something like SDSL, and even then there isn't a guarantee that you will get what you ordered.

The best way that I know of to test such a network is to actually set up the equipment needed at the CDN side of things. You can use WAN simulators to come up with simulated lines speeds similar to what will be used at the customer end. You should be able to inject different types of line problems/noise to see how that affects the transfer of the content you want to distribute. Just to be on the safe side, in addition to a simulator, you should also put as much of the real equipment together as you can - this will give you the best idea of how things will work when you actually go into production. This may get kind of pricey, however, if you don't know which vendor you will select for the equipment. One thing to ask prospective vendors is if they can give you access to their demo lab (most vendors now have one), although this could also mean travel expenses if you have to get to the lab. However, it would would let you try things out before having to buy the equipment and finding out you don't have the right pieces.

Depending on how many customers you anticipate being able to serve if you go forward with your CDN service, you might want to consider using OC3s instead of DS3s. It isn't too hard to max out a DS3 with video. If you need to use DS3s to start with because of cost reasons, also check with the various carriers to see how long it is from the time the OC3 is ordered until you have it delivered and installed. Don't be surprised if you are quoted 90 days or longer. Having a network connection maxed out for that period of time before you can get the speed increased is not somethingI think you will want to deal with. This is where getting the best help you can find before you start the ordering process will make sure you get started the right way with upgrades ordered as part of a plan instead of as a reaction.