Dial-up modems are generally only capable of a maximum bitrate of 56 kbit/s (kilobits per second) and require the full use of a telephone line—whereas broadband technologies supply at least double this speed and generally without disrupting telephone use. (It should be noted, though, that dial up is not the opposite of broadband, and is used here for practical understanding purposes only.)
Although various minimum speeds have been used in definitions of broadband, ranging up from 64 kbit/s up to 1.0 Mbit/s, the 2006 OECD report is typical in counting only download speeds equal to or faster than 256 kbit/s as broadband, and the US FCC currently defines broadband as anything above 768 kbit/s
Speeds are defined in terms of maximum download because several common consumer broadband technologies such as ADSL are "asymmetric"—supporting much slower maximum upload speeds than download.
|DS-1 (Tier 1)||1.544 Mbit/s|
|DS-3 (Tier 3)||44.736 Mbit/s|
Broadband is often called high-speed Internet, because it usually has a high rate of data transmission. In general, any connection to the customer of 256 kbit/s (0.256 Mbit/s) or more is considered broadband Internet. The International Telecommunication Union Standardization Sector (ITU-T) recommendation I.113 has defined broadband as a transmission capacity that is faster than primary rate ISDN, at 1.5 to 2 Mbit/s. The FCC definition of broadband is 200 kbit/s (0.2 Mbit/s) in one direction, and advanced broadband is at least 200 kbit/s in both directions. The Organization for Economic Co-operation and Development (OECD) has defined broadband as 256 kbit/s in at least one direction and this bit rate is the most common baseline that is marketed as "broadband" around the world. There is no specific bitrate defined by the industry, however, and "broadband" can mean lower-bitrate transmission methods. Some Internet Service Providers (ISPs) use this to their advantage in marketing lower-bitrate connections as broadband.
In practice, the advertised bandwidth is not always reliably available to the customer; ISPs often allow a greater number of subscribers than their backbone connection can handle, under the assumption that most users will not be using their full connection capacity very frequently. This aggregation strategy works more often than not, so users can typically burst to their full bandwidth most of the time; however, peer-to-peer (P2P) file sharing systems, often requiring extended durations of high bandwidth, stress these assumptions, and can cause major problems for ISPs who have excessively overbooked their capacity. For more on this topic, see traffic shaping. As takeup for these introductory products increases, telcos are starting to offer higher bit rate services. For existing connections, this most of the time simply involves reconfiguring the existing equipment at each end of the connection.
As the bandwidth delivered to end users increases, the market expects that video on demand services streamed over the Internet will become more popular, though at the present time such services generally require specialized networks. The data rates on most broadband services still do not suffice to provide good quality video, as MPEG-2 video requires about 6 Mbit/s for good results. Adequate video for some purposes becomes possible at lower data rates, with rates of 768 kbit/s and 384 kbit/s used for some video conferencing applications, and rates as low as 100 kbit/s used for videophones using H.264/MPEG-4 AVC. The MPEG-4 format delivers high-quality video at 2 Mbit/s, at the high end of cable modem and ADSL performance.
Increased bandwidth has already made an impact on newsgroups: postings to groups such as alt.binaries.* have grown from JPEG files to entire CD and DVD images. According to NTL, the level of traffic on their network increased from a daily inbound news feed of 150 gigabytes of data per day and 1 terabyte of data out each day in 2001 to 500 gigabytes of data inbound and over 4 terabytes out each day in 2002.
Diamond and other vendors had created dual phone line modems with bonding capability. The speed of dual line modems is faster than 90 kbit/s. To use this modem, the ISP should support line bonding. The Internet and phone charge will be twice the ordinary dial-up charge.
Load balancing takes two internet connections and feeds them into your network as one double speed, more resilient internet connection. By choosing two independent internet providers the load balancing hardware will automatically use the line with least load which means should one line fail, the second one automatically takes up the slack.
A basic rate ISDN line (known as ISDN-BRI) is an ISDN line with 2 data "bearer" channels (DS0 - 64 kbit/s each). Using ISDN terminal adapters (erroneously called modems), it is possible to bond together 2 or more separate ISDN-BRI lines to reach speeds of 256 kbit/s or more. The ISDN channel bonding technology has been used for video conference applications and high-speed data transmission.
Primary rate ISDN, known as ISDN-PRI, is an ISDN line with 23 DS0 channels and total speed of 1,544 kbit/s (US standard). ISDN E1 (European standard) line is an ISDN lines with 30 DS0 channels and total speed of 2,048 kbit/s. Because ISDN is a telephone-based product, a lot of the terminology and physical aspects of the line are shared by the ISDN-PRI used for voice services. An ISDN line can therefore be "provisioned" for voice or data and many different options, depending on the equipment being used at any particular installation, and depending on the offerings of the telephone company's central office switch. Most ISDN-PRI's are used for telephone voice communication using large PBX systems, rather than for data. One obvious exception is that ISPs usually have ISDN-PRI's for handling ISDN data and modem calls.
It is mainly of historical interest that many of the earlier ISDN data lines used 56 kbit/s rather than 64 kbit/s "B" channels of data. This caused ISDN-BRI to be offered at both 128 kbit/s and 112 kbit/s rates, depending on the central office's switching equipment.
Since a T-1 was originally conceived for voice transmission, and voice T-1's are still widely used in businesses, it can be confusing to the uninitiated subscriber. It is often best to refer to the type of T-1 being considered, using the appropriate "data" or "voice" prefix to differentiate between the two. A voice T-1 would terminate at a phone company's central office (CO) for connection to the PSTN; a data T-1 terminates at a point of presence (POP) or data center. The T-1 line which is between a customer's premises and the POP or CO is called the local loop. The owner of the local loop need not be the owner of the network at the POP where your T-1 connects to the Internet, and so a T-1 subscriber may have contracts with these two organizations separately.
The nomenclature for a T-1 varies widely, cited in some circles a DS-1, a T1.5, a T1, or a DS1. Some of these try to distinguish amongst the different aspects of the line, considering the data standard a DS-1, and the physical structure of the trunk line a T-1 or T-1.5. They are also called leased lines, but that terminology is usually for data speeds under 1.5 Mbit/s. At times, a T-1 can be included in the term "leased line" or excluded from it. Whatever it is called, it is inherently related to other high-speed access methods, which include T-3, SONET OC-3, and other T-carrier and Optical Carriers. Additionally, a T-1 might be aggregated with more than one T-1, producing an nxT-1, such as 4xT-1 which has exactly 4 times the bandwidth of a T-1.
When a T-1 is installed, there are a number of choices to be made: in the carrier chosen, the location of the demarcation point, the type of channel service unit (CSU) or data service unit (DSU) used, the WAN IP router used, the types of speeds chosen, etc. Specialized WAN routers are used with T-1 lines that route Internet or VPN data onto the T-1 line from the subscriber's packet-based (TCP/IP) network using customer premises equipment (CPE). The CPE typical consists of a CSU/DSU that converts the DS-1 data stream of the T-1 to a TCP/IP packet data stream for use in the customer's Ethernet LAN. It is noteworthy that many T-1 providers optionally maintain and/or sell the CPE as part of the service contract, which can affect the demarcation point and the ownership of the router, CSU, or DSU.
Although a T-1 has a maximum of 1.544 Mbit/s, a fractional T-1 might be offered which only uses an integer multiple of 128 kbit/s for bandwidth. In this manner, a customer might only purchase 1/12th or 1/3 of a T-1, which would be 128 kbit/s and 512 kbit/s, respectively.
However, in certain locations, true Ethernet broadband access might be available. This would most commonly be the case at a POP or a data center, and not at a typical residence or business. When Ethernet Internet access is offered, it could be fiber-optic or copper twisted pair, and the speed will conform to standard Ethernet speeds of up to 10 Gbit/s. The primary advantage is that no special hardware is needed for Ethernet. Ethernet also has a very low latency.
Several rural broadband solutions exist, though each has its own pitfalls and limitations. Some choices are better than others, but are dependent on how proactive the local phone company is about upgrading their rural technology.
Wireless Internet Service Provider (WISPs) are rapidly becoming a popular broadband option for rural areas.
Satellite Internet also has a high latency problem caused by the signal having to travel 35,000 km (22,000 miles) out into space to the satellite and back to Earth again. The signal delay can be as much as 500 milliseconds to 900 milliseconds, which makes this service unsuitable for applications requiring real-time user input such as certain multiplayer Internet games and first-person shooters played over the connection. Despite this, it is still possible for many games to be played, but the scope is limited to real-time strategy or turn-based games. The functionality of live interactive access to a distant computer can also be subject to the problems caused by high latency. These problems are more than tolerable for just basic email access and web browsing and in most cases are barely noticeable.
There is no simple way to get around this problem. The delay is primarily due to the speed of light being only 300,000 km/second (186,000 miles per second). Even if all other signaling delays could be eliminated it still takes the electromagnetic wave 233 milliseconds to travel from ground to the satellite and back to the ground, a total of 70,000 km (44,000 miles) to travel from the user to the satellite company.
Since the satellite is usually being used for two-way communications, the total distance increases to 140,000 km (88,000 miles), which takes a radio wave 466 ms to travel. Factoring in normal delays from other network sources gives a typical connection latency of 500-700 ms. This is far worse latency than even most dial-up modem users' experience, at typically only 150-200 ms total latency.
Most satellite Internet providers also have a FAP (Fair Access Policy). Perhaps one of the largest disadvantages of satellite Internet, these FAPs usually throttle a user's throughput to dial-up speeds after a certain "invisible wall" is hit (usually around 200 MB a day). This FAP usually lasts for 24 hours after the wall is hit, and a user's throughput is restored to whatever tier they paid for. This makes bandwidth-intensive activities nearly impossible to complete in a reasonable amount of time (examples include P2P and newsgroup binary downloading).
These can give broadband access to the Internet, with a cell phone, with Cardbus, ExpressCard, or USB cellular modems, or with cellular broadband routers, which allow more than one computer to be connected to the Internet using one cellular connection.
Broadband over power lines (BPL), also known as Power line communication, has developed faster in Europe than in the US due to a historical difference in power system design philosophies. Nearly all large power grids transmit power at high voltages in order to reduce transmission losses, then near the customer use step-down transformers to reduce the voltage. Since BPL signals cannot readily pass through transformers, repeaters must be attached to the transformers. In the US, it is common for a small transformer hung from a utility pole to service a single house. In Europe, it is more common for a somewhat larger transformer to service 10 or 100 houses. For delivering power to customers, this difference in design makes little difference, but it means delivering BPL over the power grid of a typical US city will require an order of magnitude more repeaters than would be required in a comparable European city.
The second major issue is signal strength and operating frequency. The system is expected to use frequencies in the 10 to 30 MHz range, which has been used for decades by licensed amateur radio operators, as well as international shortwave broadcasters and a variety of communications systems (military, aeronautical, etc.). Power lines are unshielded and will act as transmitters for the signals they carry, and have the potential to completely wipe out the usefulness of the 10 to 30 MHz range for shortwave communications purposes.
Traditional 802.11b was licensed for omnidirectional service spanning only 100-150 meters (300-500 ft). By focusing the signal down to a narrow beam with a Yagi antenna it can instead operate reliably over a distance of many miles.
Rural Wireless-ISP installations are typically not commercial in nature and are instead a patchwork of systems built up by hobbyists mounting antennas on radio masts and towers, agricultural storage silos, very tall trees, or whatever other tall objects are available. There are currently a number of companies that provide this service. A wireless Internet access provider map for USA is publicly available for WISPS.
In the end, the disadvantages outweighed the advantages and the glut of fiberoptic capacity that ensued following the collapse of the Internet bubble drove the cost of transmission so low that an ancillary service such as this was unnecessary, and the company folded at the end of 2005. The partner television stations as well as over 500 additional television stations not part of the iBlast Network continue to transmit separate digital signals as mandated by the Telecommunications Act of 1996.
For ISPs who are bandwidth limited, the "all you can eat" model may become unsustainable as demand for bandwidth increases. Fixed costs represent 80-90% of the cost of providing broadband service, and although most ISPs keep their cost secret, the total cost (January 2008) is estimated to be about $0.10 per gigabyte. Currently some ISPs estimate that about 5% of users consume about 50% of the total bandwidth
Some ISPs have begun experimenting with usage-based pricing, notably a Time Warner test in Beaumont, Texas. Bell Canada has imposed bandwidth caps on customers, with pricing ranging from $1 to $7.50 per gigabyte for usage over certain limits. For comparison, note that a typical standard-definition movie is 700MB-1.2GB, while a high-definition movie is 6GB-12GB. This could conceivably result in a charge of $90 to view a movie.
An often overlooked analysis when choosing an internet provider is comparing the different DSL and cable internet services at the plan level. Doing so will ensure that consumers do not overpay for speed they will not utilize. While this comparison can be cumbersome, there are resources like http://highspeed-internet-providers.com that assist in the decision making process.