Monday, February 28, 2011

Shielded Twisted-Pair (STP) | Types of Communications Media

Shielded twisted-pair (STP) cabling was first made popular by IBM when it introduced type classification for data cabling. Though more expensive to purchase and install than UTP, STP offers some distinct advantages. The current ANSI/TIA-568-C cabling standard recognizes IBM Type 1A horizontal cable, which supports frequency rates of up to 300MHz, but does not recommend it for new installations. STP cable is less susceptible to outside electromagnetic interference (EMI) than UTP cabling because all cable pairs are well shielded.

Not All UTP Is Created Equal!
Though two cables may look identical, their supported data rates can be dramatically different. Older UTP cables that were installed to support telephone systems may not even support 10Base-T Ethernet. The ANSI/TIA-568-C standard helps consumers choose the right cable (and components) for their application. The standard has been updated over the years and currently defines four categories of UTP cable: Categories 3, 5e, 6, and 6A. Here is a brief rundown of categories past and present:
  • Category 1 (not defined by ANSI/TIA-568-C) This type of cable usually supports frequencies of less than 1MHz. Common applications include analog voice telephone systems. It was never included in any version of the 568 standard.

  • Category 2 (not defined by ANSI/TIA-568-C) This cable type supports frequencies of up to 4MHz. It's not commonly installed, except in installations that use twisted-pair ARCnet and Apple LocalTalk networks. Its requirements are based on the original, proprietary IBM Cabling System specification. It was never included in any version of the 568 standard.

  • Category 3 (recognized cable type in ANSI/TIA-568-C) This type of cable supports data rates up to 16MHz. This cable was the most common variety of UTP for a number of years starting in the late 1980s. Common applications include 4Mbps UTP Token Ring, 10Base-T Ethernet, 100Base-T4, and digital and analog telephone systems. Its inclusion in the ANSI/TIA-568-C standard is for voice applications.

  • Category 4 (not defined by ANSI/TIA-568-C) Cable belonging to Category 4 was designed to support frequencies of up to 20MHz, specifically in response to a need for a UTP solution for 16Mbps Token Ring LANs. It was quickly replaced in the market when Category 5 was developed, as Category 5 gives five times the bandwidth with only a small increment in price. Category 4 was a recognized cable in the 568-A Standard, but was dropped from ANSI/TIA/EIA-568-B and also does not appear in ANSI/TIA-568-C.

  • Category 5 (was included in ANSI/TIA/EIA-568-B for informative purposes only) Category 5 was the most common cable installed, until new installations began to use an enhanced version. It may still be the cable type most in use because it was the cable of choice during the huge infrastructure boom of the 1990s. It was designed to support frequencies of up to 100MHz. Applications include 100Base-TX, FDDI over copper, 155Mbps ATM over UTP, and, thanks to sophisticated encoding techniques, 1000Base-T Ethernet. To support 1000Base-T applications, the installed cabling system had to pass performance tests specified by TSB-95 (TSB-95 was a Telecommunications Systems Bulletin issued in support of ANSI/TIA/EIA-568-A, which defines additional test parameters). It is no longer a recognized cable type per the ANSI/TIA-568-C standard, but for historical reference purposes, Category 5 requirements, including those taken from TSB-95, are specified in ANSI/TIA-568-C.2.

  • Category 5e (recognized cable type in ANSI/TIA-568-C) Category 5e (enhanced Category 5) was introduced with the TIA/EIA-568-A-5 addendum of the cabling standard. Even though it has the same rated bandwidth as Category 5, that is, 100MHz, additional performance criteria and a tighter transmission test requirement make it more suitable for high-speed applications such as Gigabit Ethernet. Applications are the same as those for Category 5 cabling. It is now the minimum recognized cable category for data transmission in ANSI/TIA-568-C.

  • Category 6 (recognized cable type in ANSI/TIA-568-C) Category 6 cabling was officially recognized with the publication of an addition to ANSI/TIA/EIA-568-B in June 2002. In addition to more stringent performance requirements as compared to Category 5e, it extends the usable bandwidth to 250MHz. Its intended use is for Gigabit Ethernet and other future high-speed transmission rates. Successful application of Category 6 cabling requires closely matched components in all parts of the transmission channel, that is, patch cords, connectors, and cable.

  • Category 6A or Augmented Category 6 (recognized cable type in ANSI/TIA-568-C) Category 6A cabling was officially recognized with the publication of ANSI/TIA/EIA-568-B.2-10 in February 2008. In addition to more stringent performance requirements as compared to Category 6, it extends the usable bandwidth to 500MHz. Its intended use is for 10 Gigabit Ethernet. Like Category 6, successful application of Category 6A cabling requires closely matched components in all parts of the transmission channel, that is, patch cords, connectors, and cable. 

  • Category 7 (recognized cable type in ISO 11801) Category 7 is an ISO/IEC category suitable for transmission frequencies up to 1GHz. It is widely used in Europe and is gaining some popularity in the United States. It is not presently recognized in ANSI/TIA-568-C.

Some STP cabling, such as IBM types 1 and 1A cable, uses a woven copper-braided shield, which provides considerable protection against EMI. Inside the woven copper shield, STP consists of twisted pairs of wire (usually two pairs) wrapped in a foil shield. Some STP cables have only the foil shield around the wire pairs.

New Nomenclature for Twisted-Pair Cables
TIA is addressing the potentially confusing nomenclature for different types of twisted-pair cables:
  • Shielded twisted-pair (STP) will be called U/FTP.
  • Screened twisted-pair (ScTP or FTP) will be called F/UTP.
  • Category 7 screened shielded twisted-pair (S/STP or S/FTP) will be called ScFTP.

Figure 1 shows a typical STP cable. In the IBM design, the wire used in STP cable is 22 AWG (just a little larger than the 24 AWG wire used by typical UTP LAN cables) and has a nominal impedance of 150 ohms, but category versions can have a nominal impedance of 100 ohms.

Figure 1: STPcable

Constructions of STP in 24 AWG, identical in copper conductor size to UTP cables, are more commonly used today.

Simply installing STP cabling does not guarantee you will improve a cable's immunity to EMI or reduce the emissions from the cable. Several critical conditions must be met to achieve good shield performance:
  • The shield must be electrically continuous along the whole link.
  • All components in the link must be shielded. No UTP patch cords can be used.
  • The shield must fully enclose the pair, and the overall shield must fully enclose the core. Any gap in the shield covering is a source of EMI leakage.
  • The shield must be grounded at both ends of the link, and the building grounding system must conform to grounding standards (such as J-STD-607-A).
If even one of these conditions is not satisfied, shield performance will be badly degraded. For example, tests have shown that if the shield continuity is broken, the emissions from a shielded cabling system increase by 20dB on the average.

Friday, February 25, 2011

The Importance of Reliable Cabling

We cannot stress enough the importance of reliable cabling. Two recent studies vindicated our evangelical approach to data cabling. The studies showed:
  • Data cabling typically accounts for less than 10 percent of the total cost of the network infrastructure.
  • The life span of the typical cabling system is upward of 16 years. Cabling is likely the second most long-lived asset you have (the first being the shell of the building).
  • Nearly 70 percent of all network-related problems are due to poor cabling techniques and cable-component problems.

If you have installed the proper category or grade of cable, the majority of cabling problems will usually be related to patch cables, connectors, and termination techniques. The permanent portion of the cable (the part in the wall) will not likely be a problem unless it was damaged during installation.

Of course, these were facts that we already knew from our own experiences. We have spent countless hours troubleshooting cabling systems that were nonstandard, badly designed, poorly documented, and shoddily installed. We have seen many dollars wasted on the installation of additional cabling and cabling infrastructure support that should have been part of the original installation.

Regardless of how you look at it, cabling is the foundation of your network. It must be reliable!

The Cost of Poor Cabling
The costs that result from poorly planned and poorly implemented cabling systems can be staggering. One company that moved into a new datacenter space used the existing cabling, which was supposed to be Category 5e cable. Almost immediately, 10 Gigabit Ethernet network users reported intermittent problems.
These problems included exceptionally slow access times when reading email, saving documents, and using the sales database. Other users reported that applications running under Windows XP and Windows Vista were locking up, which often caused them to have to reboot their PC.

After many months of network annoyances, the company finally had the cable runs tested. Many cables did not even meet the minimum requirements of a Category 5e installation, and other cabling runs were installed and terminated poorly.

Often, network managers mistakenly assume that data cabling either works or it does not, with no in-between. Cabling can cause intermittent problems.

Is the Cabling to Blame?
Can faulty cabling cause the type of intermittent problems that the aforementioned company experienced? Contrary to popular opinion, it certainly can. In addition to being vulnerable to outside interference from electric motors, fluorescent lighting, elevators, cell phones, copiers, and microwave ovens, faulty cabling can lead to intermittent problems for other reasons.

These reasons usually pertain to substandard components (patch panels, connectors, and cable) and poor installation techniques, and they can subtly cause dropped or incomplete packets. These lost packets cause the network adapters to have to time out and retransmit the data.

Robert Metcalfe (inventor of Ethernet, founder of 3Com, columnist for InfoWorld, and industry pundit) helped coin the term drop-rate magnification. Drop-rate magnification describes the high degree of network problems caused by dropping a few packets. Metcalfe estimates that a 1 percent drop in Ethernet packets can correlate to an 80 percent drop in throughput. Modern network protocols that send multiple packets and expect only a single acknowledgement are especially susceptible to drop-rate magnification, as a single dropped packet may cause an entire stream of packets to be retransmitted.

Dropped packets (as opposed to packet collisions) are more difficult to detect because they are "lost" on the wire. When data is lost on the wire, the data is transmitted properly but, due to problems with the cabling, the data never arrives at the destination or it arrives in an incomplete format.

Sunday, February 20, 2011

The Legacy of Proprietary Cabling Systems

Early cabling systems were unstructured, proprietary, and often worked only with a specific vendor's equipment. They were designed and installed for mainframes and were a combination of thicknet cable, twinax cable, and terminal cable (RS-232). Because no cabling standards existed, an MIS director simply had to ask the vendor which cable type should be run for a specific type of host or terminal. Frequently, though, vendor-specific cabling caused problems due to lack of flexibility. Unfortunately, the legacy of early cabling still lingers in many places.

PC LANs came on the scene in the mid-1980s; these systems usually consisted of thicknet cable, thinnet cable, or some combination of the two. These cabling systems were also limited to only certain types of hosts and network nodes.

As PC LANs became popular, some companies demonstrated the very extremes of data cabling. Looking back, it's surprising to think that the ceilings, walls, and floor trenches could hold all the cable necessary to provide connectivity to each system. As one company prepared to install a 1,000-node PC LAN, they were shocked to find all the different types of cabling systems needed. Each system was wired to a different wiring closet or computer room and included the following:

  • Wang dual coaxial cable for Wang word processing terminals
  • IBM twinax cable for IBM 5250 terminals
  • Twisted-pair cable containing one or two pairs, used by the digital phone system
  • Thick Ethernet from the DEC VAX to terminal servers
  • RS-232 cable to wiring closets connecting to DEC VAX terminal servers
  • RS-232 cable from certain secretarial workstations to a proprietary NBI word processing system
  • Coaxial cables connecting a handful of PCs to a single Novell NetWare server
Some users had two or three different types of terminals sitting on their desks and, consequently, two or three different types of wall plates in their offices or cubicles. Due to the cost of cabling each location, the locations that needed certain terminal types were the only ones that had cables that supported those terminals. If users moved—and they frequently did—new cables often had to be pulled.

The new LAN was based on a twisted-pair Ethernet system that used unshielded twisted-pair cabling called SynOptics LattisNet, which was a precursor to the 10Base-T standards. Due to budget considerations, when the LAN cabling was installed, this company often used spare pairs in the existing phone cables. When extra pairs were not available, additional cable was installed. Networking standards such as 10Base-T were but a twinkle in the IEEE's (Institute of Electrical and Electronics Engineers) eye, and guidelines such as the ANSI/TIA/EIA-568 series of cabling standards were not yet formulated. Companies deploying twisted-pair LANs had little guidance, to say the least.

Much of the cable that was used at this company was sub–Category 3, meaning that it did not meet minimum Category 3 performance requirements. Unfortunately, because the cabling was not even Category 3, once the 10Base-T specification was approved, many of the installed cables would not support 10Base-T cards on most of the network. So three years into this company's network deployments, it had to rewire much of its building.

KEY TERM: aplication 
Often you will see the term application used when referring to cabling. If you are like us, you think of an application as a software program that runs on your computer. However, when discussing cabling infrastructures, an application is the technology that will take advantage of the cabling system. Applications include telephone systems (analog voice and digital voice), Ethernet, Token Ring, ATM, ISDN, and RS-232.

Proprietary Cabling is a Thing of the Past
The company discussed had at least seven different types of cables running through the walls, floors, and ceilings. Each cable met only the standards dictated by the vendor that required that particular cable type.
As early as 1988, the computer and telecommunications industry yearned for a versatile standard that would define cabling systems and make the practices used to build these cable systems consistent. Many vendors defined their own standards for various components of a cabling system.

The Need for a Comprehensive Standard
Twisted-pair cabling in the late 1980s and early 1990s was often installed to support digital or analog telephone systems. Early twisted-pair cabling (Level 1 or Level 2) often proved marginal or insufficient for supporting the higher frequencies and data rates required for network applications such as Ethernet and Token Ring. Even when the cabling did marginally support higher speeds of data transfer (10Mbps), the connecting hardware and installation methods were often still stuck in the "voice" age, which meant that connectors, wall plates, and patch panels were designed to support voice applications only.

The original Anixter Cables Performance Levels document only described performance standards for cables. A more comprehensive standard had to be developed to outline not only the types of cables that should be used but also the standards for deployment, connectors, patch panels, and more.

A consortium of telecommunications vendors and consultants worked in conjunction with the American National Standards Institute (ANSI), Electronic Industries Alliance (EIA), and the Telecommunications Industry Association (TIA) to create a Standard originally known as the Commercial Building Telecommunications Cabling Standard, or ANSI/TIA/EIA-568-1991. This standard has been revised and updated several times. In 1995, it was published as ANSI/TIA/EIA-568-A, or just TIA/EIA-568-A. In subsequent years, TIA/EIA-568-A was updated with a series of addendums. For example, TIA/EIA-568-A-5 covered requirements for enhanced Category 5 (Category 5e), which had evolved in the marketplace before a full revision of the standard could be published. A completely updated version of this standard was released as ANSI/TIA/EIA-568-B in May 2001. At the time of this writing, a new standard is about to be released, called ANSI/TIA-568-C; 

The IEEE maintains the industry standards for Ethernet protocols (or applications). This is part of the 802.3 series of standards and includes applications such as 1000Base-T, 1000Base-SX, 10GBase-T, and 10GBase-SR.

The structured cabling market is estimated to be worth approximately $5 billion worldwide (according to the Building Services Research and Information Association [BSRIA]), due in part to the effective implementation of nationally recognized standards.

Thursday, February 10, 2011

The Golden Rules of Data Cabling

Listing our own golden rules of data cabling is a great way to start this chapter and the book. If your cabling is not designed and installed properly, you will have problems that you can't even imagine. Using our experience, we've become cabling evangelists, spreading the good news of proper cabling. What follows is our list of rules to consider when planning structured-cabling systems:

  • Networks never get smaller or less complicated.
  • Build one cabling system that will accommodate voice and data.
  • Always install more cabling than you currently require. Those extra outlets will come in handy someday.
  • Use structured-cabling standards when building a new cabling system. Avoid anything proprietary!
  • Quality counts! Use high-quality cabling and cabling components. Cabling is the foundation of your network; if the cabling fails, nothing else will matter. For a given grade or category of cabling, you'll see a range of pricing, but the highest prices don't necessarily mean the highest quality. Buy based on the manufacturer's reputation and proven performance, not the price.
  • Don't scrimp on installation costs. Even quality components and cable must be installed correctly; poor workmanship has trashed more than one cabling installation.
  • Plan for higher-speed technologies than are commonly available today. Just because 1000Base-T Ethernet seems unnecessary today does not mean it won't be a requirement in five years.
  • Documentation, although dull, is a necessary evil that should be taken care of while you're setting up the cabling system. If you wait, more pressing concerns may cause you to ignore it.