Just a few years ago, a newly initiated network administrator in a small company had enormous trouble arranging for a full-time connection to the Internet. Times have changed, and many choices for affordable, full-time network connections are available to everyone from the home user to the largest enterprise. Newly anointed network administrators will want to know what they are talking about when they go to make that first Internet connection purchase. In this Daily Drill Down, I’ll look at some of the options available, the hardware required, the implementation costs, and monthly charges.
CLECs and LECs
An option that will be available in most parts of the United States is using a Competitive Local Exchange Carrier (CLEC), a company that competes with the already established local telephone business by providing its own network and switching, to provide your Internet connection instead of dealing with the telco (LEC), as most companies have done for years. Although a CLEC can sometimes be cheaper than the telco, it comes with a caveat. Some CLECs do nothing more than sublet the copper from the telco. While this may not sound like a big deal, it can lead to a few more headaches when you have line problems. Basically, the CLEC can only troubleshoot so far before it’ll have to put a call in to the telco to have it look at the line. In this case, you aren’t much better off than just dealing with the telco directly; you now have multiple parties that have to work together when you have line problems. If the CLEC has its own copper or fiber and has better pricing than the telco, check into how reliable their service is with some of the CLEC’s customers—you may find a real deal.
Point-to-point and frame relay
This has long been the way most companies have connected to the Internet. You’ll find either some type of point-to-point connection (also known as a private line) between your company and the ISP or something known as frame relay. There is a main difference between the two. The point-to-point circuit is a direct connection between you and the ISP’s POP (point of presence, or where you connect to the equipment that connects to the ISP’s network). Meanwhile, a frame relay connection connects you to a cloud, or virtual network, that allows traffic (to and from your network) to travel on different routes between you and the ISP.
With a frame relay connection, you have to know some additional terminology to get the type of connection you want. The two most important terms are committed information rate (CIR) and burst. CIR is the minimum line speed that the telco or CLEC has to provide to you according to your contract. Burst is the maximum line speed that they can provide you with if (if being the key word) the bandwidth is available at the time you need it. Having burst on a frame relay circuit is a catch-22 situation. If the capacity on the frame relay network is available, you’ll be able to use it on an on-demand basis. Here’s the catch: You’ll incur extra charges each time you start using bandwidth beyond what your contract states you’re entitled to. The extra charge will come in at the CIR rate and up to the amount specified as the burstable bandwidth capacity. This can lead to a nasty surprise when you get the monthly bill.
Keep in mind that some ISPs and telcos will bet that you won’t always use the additional capacity in the burst speed range, so they’ll have additional subscribers also using the same frame relay network that you are on. This can lead to oversubscription—subscribers attempting to use more bandwidth than is available. This is an important issue to bring up with the ISP and the telco. You want to get an idea of whether you’ll be able to get burstable bandwidth or if you’ll even get the line speed or bandwidth that you have contracted for.
With a point-to-point connection, part of what will make the circuit more costly than its frame relay cousin is your distance from the central office (CO) where the line terminates and how far away you are from the ISP’s POP. Although you can expect this type of line to give you better performance, it can also be more expensive. If you do go with this type of connection, make sure when the line is ordered either by you or the ISP that Extended Super Frame (ESF) is specified as the frame type to be used on the line. Better than standard Super Frame (which consists of 12 frames with 192 bits each vs. ESF’s 24 frames with 192 bits each), ESF allows for more diagnostics to be run on the line without taking the circuit out of service before beginning any testing. At some point, the circuit will have to be taken out of service for testing but at least this way, you can avoid as much downtime as possible.
Regardless of which type of line you choose, the next piece of gear you’ll need is a CSU/DSU. This is basically a modem for leased-line applications. It’s the one piece of equipment that you’ll need to connect the leased line to the router. It can either be an external unit or take a slot in a router. My preference is to use an external unit for two reasons. With the external CSU/DSU, you have one or more status lights that indicate whether the data line you’re connecting to is up and possibly whether an alarm condition exists on the line that could cause a problem. The other reason is that, in the unlikely event that you suffer a lightning strike, you have a slightly better chance of losing just the CSU/DSU instead of losing the router as well.
You can choose from several good brands of CSU/DSUs. I am a big fan of Vanguard’s line of products because of its reliability and support. I recently started using a CSU/DSU from ADTRAN. This line of leased-line modems has a feature that should make it attractive to a lot of folks; it has intelligence built into it that learns how the data circuit you’re connected to is configured—it adjusts itself automatically to the correct settings.
Another choice will come when you have more than one T-1 connecting your company to the Internet. As you shop for CSU/DSUs, you’ll find two versions of just about every model you look at. One will be a standalone unit, which can sit on a shelf and will handle one data circuit. The other version is a rack-mount type that will rest in a 19” equipment rack; you insert a card in it for each T-1 that you will be using. When connecting one or two T-1s, the rack-mount option can be expensive due to the cost of the cage for the CSU/DSU cards. This is where ADTRAN has an interesting offer. It has a small shelf that bolts into your 19” equipment rack and can handle two of the standalone ADTRAN CSU/DSUs.
As you consider the different ISPs that want your business, you’ll find that some of them offer a package deal for service that includes the CSU/DSU and router. They’ll probably offer a purchase and a rental option. If this is your first experience with a router or you just don’t want to be hassled by the setup and configuration process, the rental option may be the best way to go. With that option, if you can ping the router but can’t go further, the problem is probably within the router or at the ISP, and then it becomes someone else’s headache. Also, if the router fails or you don’t want to get inside the router to program it or add a feature, this is also the ISP’s responsibility.
Several different flavors of xDSL are available. Asymmetric digital subscriber line (ADSL) is implemented in varying ways, depending on the equipment the DSL provider has chosen to use at the CO. The real plus with ADSL is that it can usually run on top of the existing pair of copper wires that your phone service uses. Without getting into the nitty-gritty of how ADSL works, it basically uses a different set of frequencies or tones from those that are required for voice service. Most ADSL providers will send out a self-installation kit and will also include some line filters for you to place at the outlets at your office so that the DSL signal won’t interfere with other devices, such as fax machines.
Different line speeds are available with DSL. The speeds available to your specific location will be dependent in great part on how far away you are from the CO. Simply put: the greater the distance, the slower the connection.
I’ve dealt with several telcos around the country and have found that they seem to be challenged by getting the DSL service installed and keeping it running. The biggest problems I’ve seen with installation come down to two things: voice coils and bridge taps. The voice coils are actual loading coils placed on the voice line to make the voice service more reliable and better sounding. While that is good, it can cause a real problem in getting DSL to work reliably, if at all. The bridge taps aren’t really a line splice but are a way for telcos to reroute a voice circuit to another location without rerunning the cable. Bridge taps change the impedance (electrical characteristics) of the line and can cause problems with the DSL service running correctly.
One of the first things that the telco or DSL provider will do before initiating the server order to install the DSL service will be to check to see how far you are from the CO. Some telcos won’t install service if you are over 16,000 feet from the CO. I have seen DSL service work at distances of up to 20,000 feet, but your connection may end up so slow that it won’t warrant the effort of subscription.
The available line speeds with cable modems are faster, and the cost for service is usually less than for the xDSL offerings. The downside to this type of service is that the cable company in your area may not be able to offer this to businesses until all demanding consumers have been connected to this technology.
What you’ll have in this case is a cable modem that uses an F connector to connect to the service from your cable provider. There will also be an RJ-45 connector that will connect to your network. Most service offerings will provide a single IP address that will be assigned to a computer that connects to the cable modem. To connect your network to this, you’ll need some type of DSL router (I have used the Linksys router with good results) that will be programmed with the system ID as provided by the Internet cable service provider to make sure that only authorized computers are on the network. You’ll definitely want some type of firewall on this connection to protect the systems that you’ll be connecting to the Internet from unwanted intruders.
Depending on the cable company offering the Internet service, you can expect speed somewhere around the range of a T-1. While some limitations exist in what you can run on the cable modem Internet connection (some systems won’t allow you to run either NNTP [Usenet news feeds] or DNS servers), you may find this to be an economical alternative to having a conventional T-1 data line.
Once an option for companies that had to have connectivity at any cost and weren’t reachable by other means, this option is now very affordable (less than $100 a month) and, unlike earlier offerings, is handled entirely over the satellite; no local dial-up modem is required for the uplink connection. You can expect speeds comparable to ADSL. The engineering folks at Starband told me that you can expect around 500 Kbps (about one-third of a T-1) in download speed and approximately 256 Kbps in upload speed. This is subject to variation based on the number of people using the system at the same time and inclement weather between you and the satellite. The hardware required at your end of the connection will run around $300, with an installation charge of $100 to $150. This option currently requires that an approved installer do the installation. If you try to do it yourself, you won’t have any warranty on the equipment.
This option, while higher-priced than the cable modem option, does provide an alternative to being restricted to dial-up only for Internet access at remote offices. Due to the low monthly cost of satellite Internet access, it’s an option for a backup Internet connection at those times when your faster modes of access are unavailable.
What I’ve discussed here are connectivity options that are available in most parts of the country. Some wireless Internet providers are using spread spectrum radio technologies that are good only in major metropolitan areas; these should be considered as an option when available. While your connectivity needs may require that you use a conventional T-1 data line to connect to the Internet, you might want to consider having one of the other options discussed here as a backup to keep things running in the event that your primary connection isn’t available.