It seems much of what used to require a BNC or D-type connector is now handled by twisted pair and a modular connector. Facilities are less coax intensive and more twisted pair focused. What I was calling an RJ45 is more properly termed an “8P8C” connector.
Time to update my knowledge of CAT x cables and connectors. As I reviewed some literature, I became lost in the ‘weeds’ of product pitches and terminology. That’s when Alan Frank, senior systems engineer at OOBAXS came to my rescue.
As I reviewed some literature, I became lost in the ‘weeds’ of product pitches and terminology. That’s when Alan Frank, senior systems engineer at OOBAXS came to my rescue.
To share his expertise, the next couple of blog posts will be handled by Alan. He will provide a tutorial on professional applications of Gigabit Ethernet, looking at connectors, bandwidth and KVM solutions. Through these articles, Alan will offer some first-hand insight into keeping your data (even when that means video/audio) moving properly through your facility.
Editor
--------------------------------------------------------------------------------------------------------------------
An Introduction to Network Cabling 101 – Using Copper UTP for Gigabit Ethernet
By Alan M Frank, Senior Systems Engineer, out-of-band access and technology strategist, OOBAXS
I grew up in the computer industry with the onset of the MAC and PC in the mid-80s. Initially trained by manufacturers and learning in the field through trial-and-error was where a lot of R&D occurred. I suppose that over time, like with any profession, we get accustomed to our ways based on our education and the proven trial-by-fire tactics that got us through… at that time!
So, we start to get set in our ways based on our experiences and routines; suddenly we learn that things have evolved...but, not necessarily with our knowledge.
The other day, while in a nearly completed broadcast machine room, several comments were made on the sterility of the cabling; how pristine things were laid out in neatly wrapped bundles of various colored cables. I found myself immediately scanning for improper bends in the cables and the use of zip-ties or hook and loop fasteners that might be so tight that the cabling was crimped or kinked. I am always concerned about this common issue, as I know through trouble-shooting networks that even minor cabling faux pas can cause problems that can be extremely hard to trace.
Kinked or mashed cables that were re-straightened, bare wires, open sheathing or “shiners”, and untwisted cables at the termination point can all cause issues that are difficult to track down and often go uncorrected.
Such installation and maintenance errors are probably just my personal pet-peeves, yet they are common and can cause tolerable, yet problematic, conditions. While the resulting errors may only affect performance minimally, they may result in a considerable waste of time when troubleshooting. There are two common installation mistakes:
o Solid-Core copper wiring connected to 8P8C connectors meant for stranded copper
o Solid-Core wire used for patch-cords
It’s a home run
An accepted practice (in spite of the fact that it is improper), is called “home runs”. This is the practice of using solid-core wire for horizontal runs with the final termination being the 8P8C adapter that are plugged straight into the network appliance or device.
This practice of cabling likely evolved from fast Ethernet deployments (pre 100BASE-TX) when the speed difference for a dropped packet wasn’t much to consider. However, with today’s high-speed and GigE networks, attention to the details is prudent; as data rates increase, errors increase exponentially.
It is typical in networking that nearly 80% of failures and errors are attributed to the physical network infrastructure. This issue hasn’t changed much over the years, we just find the faults much, much faster (and so do the end users)!
Although practical and appearing to be considerably less expensive at first, running solid-core cable directly from appliance to switch (home runs) can and does cause issues. Imagine the rigidity of one single set of solid-core copper wire, then add an additional 7 pairs. Then bundle them all together into a small group and you’ve created a pretty solid and very stationary cable group. What happens is the cable group now moves as a whole unit at the bundle, but individually at each stationary point and this often causes intermittent continuity at the connection to the networked appliance.
Solid-core vs stranded cable
These issues are compounded if the solid-core home run cabling is terminated in a stranded-wire 8P8C connector. Solid and stranded cable require different 8P8C connectors. Stranded copper connectors are usually limited to field-prepared patch cords.
In my professional opinion, solid-core wiring should be limited to horizontal runs only and then terminated on a 110-type, or better - CATx certified termination points. From there the signal should be distributed to the appliance, switch, or cross-connect using certified pre-connectorized patch cables from a known and approved manufacturer.
Face it – the factory manufacturing process is much more consistent than a technician sitting in an unfinished room on a spool with probably minimal light and no A/C. Interestingly, the person who taught me token ring cabling was color-blind!
System degradation
Initial certification of cabling can ensure a level of performance, but it is only valid at the time of the scan. Scans are typically done during the original cable installation as part of approval and acceptance of the contractor’s work. Over time, additional equipment will be added to any equipment room. The environment changes, probably getting hotter. At 95 degrees, the integrity of a typical UTP cable begins to deteriorate. Equipment and cables get bumped or moved and integrity of a cable installation will change, and so too may the system’s performance.
In a data network, these sometimes minute changes reveal themselves as an intermittent CRC, or dropped packets. Because the digital data network is self-healing it continues to transmit and recovers quickly enough that the user doesn’t notice.
I grew intolerable to this problem early on in the diagnostics and support of analog-based KVM networks, which are not self-healing. Intermittent data loss in an analog KVM solution causes display flickers and a loss of continuity resulting in a lousy user experience. Data loss also may be exhibited as sluggish mouse or keyboard commands. Video monitors may lose a single color. If too much data gets lost, the system may reboot. Not a good thing to happen going into the 6PM news.
Digital solutions
So how did I learn this and know it was something to pay attention to? My first official cabling certification came in the ‘90s from Panduit, Leviton, and Champlain. These included day-long classes for both the copper and fiber disciplines for use in high-speed networks A high-speed network (at the time, the 10BASE-T) was easier to install and expand than analog systems.
In the late ‘90’s I spent a year working with Microtest – creators of TDR and OTDR cable testers. While troubleshooting an analog KVM matrix, I found myself describing the process of identifying the problem as similar to trying to stomp out a grass fire. Just when you’ve solved one problem another pops up.
Now, back to my visit the new machine room. I couldn’t resist commenting on the organized and pristine cabling layout, and to the well-documented manner in which the cabling was deployed.
Patch cord standards
The group made the typical comments on cable fasteners, bend radius, and the use of zip-ties. Everyone acknowledged the consideration given to those practices and that everything appeared to be in order. I personally had to comment on the use of custom-built 5-inch CATx patch cables to make things appear neat and orderly.
Although these patch cables were custom-made by a well known company, they were well under the minimum specification of what a patch-cord was supposed to be – one meter. This practice occurs a lot, and so I continued to query my team to see if they had anything new to offer on this practice. The group was split down the middle: half acknowledged that they had heard of this specification and were concerned, and the other half never heard of such a practice (including the physicist in the group). Now I was really curious – was this an actual practice, or just a myth?
The reasoning behind this practice has to do with the need for the signals to be regenerated completely between connection points. Learning this information made me do an Internet query and found no support or mention for the minimum one-meter patch cord length requirement.
As long as the patch-cord is stranded and terminated with the correct 8P8C connector, and under the 25-foot limit, it can be as short as you’d like. Therefore, there is no longer any reason for messy or unorganized patch cord cross-connects.
System signal loss
What’s left now is the concern about the number of connections because that results in signal loss and meeting the facility’s general specifications.
As a rule of thumb, each connection point reduces signal strength by 2dB. With each connection, more signal is lost and eventually the system becomes more susceptible to interference. In addition to the total number of connections, overall patch length also must be considered, especially when multiple connections are required.
A typical CATx cable run averages four connections; host-to-patch-cable, patch-cable to horizontal cabling termination point, distribution point to patch-cable, and patch-cable to switch or network appliance. The signal is regenerated at the Ethernet switch and the distance limitation is restarted. Each time the signal is restarted is called a hop.
The above standards are rudimentary for the network-savvy tech. However, attention to the small details will ensure that the network in question can transition easily from today’s 100Mb/s to the 1000Mb/s speeds just around the corner.
The key to a successful transition is to build upon the physical infrastructure in order to sustain the highest bandwidth utilization networks for use in a broadcast and video environment. Of course, it is imperative to make this type of network transition in order to sustain the demands of today’s high speed networks that transport video and broadcast traffic.
Alan M Frank is the senior systems engineer, out-of-band access and technology strategist at OOBAXS (pronounced Out-Of-Band Access)
Caveat:
My narrative should be considered as a guideline for deploying a better and more consistent network infrastructure, and to increase awareness of acceptable practices that minimize infrastructure issues. It is to be considered in addition to those publicized standards for CATx. While the practices mentioned above may be considered acceptable, they may not be in your best interest over the long run. Consult the appropriate standards references or rely on a trusted professional and watch for these visible and problematic signs whenever possible.
Additional resources:
IEEE Ethernet Standards can be found at; http://standards.ieee.org/about/get/802/802.3.html
For a thorough checklist of the visible signs to watch for in your post cabling walk-through, visit: http://oobaxs.com/101checklist.
