| Cabling lifecycles and total cost of ownership |
| Thursday, 01 February 2007 00:00 | |||
|
Total cost of ownership is becoming an increasingly important factor in the purchasing decision for structured cabling systems as Carrie Higbie of Siemon explains to Data Centre Management There are several factors that must be taken into consideration when determining the category or class of cabling that will be used in a network infrastructure. This is true for both copper and fibre. Factors that must be taken into consideration are: With the IEEE 802.3an 10GBASE-T standard now ratified, performance demands on cabling infrastructures are expected to increase over the next few years. Cabling typically represents 5-7% of an overall network budget. Cabling systems, both copper and fibre, are designed to perform for 10 years, supporting 2-3 generations of active electronics. Overall lifecycle costs should be closely considered. In some instances, overall network capabilities change at a greater pace than originally expected. This can shorten the lifecycle of a cabling system. Category 4 is a good example. This cable had a very short lifecycle due to expanding network performance requirements and the capabilities of higher-performing category 5 and, eventually, category 5e. With the advent of 10GBASE-T, a higher performing category 6 cable known as Augmented Category 6 (AC6) has been introduced. So the question is posed: How do I maximize my cabling investment, and what category of cabling should I install at my facility? As standards eliminate or rescind support for cabling categories, the active equipment manufacturers will, as history shows, follow suit. There is an intricate balance between forward movement in technology and addressing the needs of legacy systems. The final cabling choices for the 10GBASE-T standard are: (a) installed legacy category 6 with a supported distance up to 55 meters, (b) augmented category 6 (UTP or F/UTP) and (c) category 7/class F, with the latter two supporting a distance up to 100 meters. It is important to note that category 5e is not being considered as suitable for 10G operations. In order to support future 10GBASE-T applications (which is likely to occur over the next 10 years) 5e systems will be replaced. You will pay fees for removal of abandoned cable (now required by fire codes and legislation in many countries) and to allow some reuse of pathway and spaces. Part of the cabling system selection process should include the cost of the cabling itself as well as other factors that contribute to the overall cost over its entire lifetime. As mentioned previously, a cabling infrastructure should last a customer 10 years and support 2-3 iterations of active equipment and applications. A costly factor in these calculations is labour, which will vary depending on geographic location. But in many locations will be the single most costly factor over the lifecycle of a cabling plant. The comparison table does not take into account after-hours installation with overtime or tracing cables if the labelling and documentation on the system was not maintained but in these situations, these costs would contribute to and increase the overall total cost of ownership. The analysis compares the total cost of ownership for a 24 channel cabling system ranging from category 5e through category 7/class F. Initial installation costs include the cost of components, installation and testing. The figures are based on average installation costs for an LSOH system. The system total cost of ownership includes the cabling and installation components at the time of original installation and costs for remediation for each cabling plant to go from today’s 10/100 applications to 1G through 10G. This result is divided by the years of useful life to achieve annualized costs. Category 7/Class F systems have a longer lifecycle, as this cable will support 10G as well as 40G, which is expected to be the next speed increase for copper. Remediation costs include downtime and lost productivity costs due to testing or recabling. Downtime costs are based on national average wages and average lost revenues from published figures. While there is a small differential in the component costs at the time of original installation, adding labour to test for additional performance parameters or removing non-compliant channels increases the total cost of ownership for the lesser performing systems over their useful life. Note the category 6 and 5e systems also have a shorter useful life so their annualized costs are higher than systems that support faster applications over time. From the point that 5e is installed to the requirements to run gigabit (including re-testing for the new power sum parameters and remediation of channels that fail), adding back the labour to remove the system for 10G, we can see that the overall total cost of ownership (total cost of installation + additional labour costs / number of years of useful life = annualized costs) is the most expensive system. The 5e system shown above is depicted in annualized form to the point in which it is removed. Obviously when a new system is installed, a new total cost of ownership calculation will apply. So for the costs shown above with 5e at 10 gigabit speeds, we have NO cabling system for our network. For the Category 6 system, the installation costs, plus testing to determine channel length and to replace channels over 55m are all added for a total cost of ownership. Again, we divide the total cost of ownership of the system by the number of years of useful life for the annualized cost. Additional labour will be needed for mitigation to support 55m, but varies by site and is not included. Examples include, allocating more space for new patch panels, increasing the number of core drills to accommodate lower fill limits, etc. The lower fill limits are necessary to mitigate AXT. No change is necessary for 10G 6 or 10G 6A F/UTP or Category 7 TERA to support 10 gigabit. The greatest disturber in 10Gigabit systems is Alien Crosstalk (AXT). In order to decrease this cable-to-cable noise one will need to either mitigate it or move to a shielded system where AXT is not a concern. The table above does not include mitigation costs for existing plants, only the testing and replacement for channels over 55m. Mitigation techniques include things such as moving to shielded patch cords, shortening lengths, unbundling cables, providing port separation at outlets and panels, etc. There is also a new augmented category 6 UTP cable that has an overall allowable diameter of 9mm. This increased diameter mitigates AXT by providing separation of conductors from one cable to the next. Compare the 9mm diameter to category 6 cable (6.35mm) and category 6A F/UTP (6.73mm). In instances where the newer larger cable is installed, conduit fill ratios and cable tray fill ratios must also be increased. For 1000 channels at 100m, the additional costs for an augmented category 6 system including pathways are in the chart above. To Shield or Not to Shield While some countries such as Germany and France have held fast to shielded systems, some countries such as the UK have traditionally embraced UTP systems outside of factory environments. There are some benefits to shielded systems in particular through increased capacity due to the reduction of disturbing noise such as AXT. Category 7/Class F systems are under a change to become Category 7A/Class FA systems capable of carrying one gigahertz as opposed to 10G 6 systems rated for 500 MHz. This cable is referred to as S/FTP or shielded/foiled twisted pairs. There is a shield around each individual pair and an overall braid shield around the entire cable. This cable has been tested as suitable for use in classified networks due to its elimination of radiated noise through the shielding and connector. The same shields create an immunity to noise that could be received into the channel. 10G 6A F/UTP is a UTP cable with an overall shield. This cable provides the second highest level of immunity to noise and has an increased data capacity over UTP systems. This means that shielded systems will carry 10G signals and are likely to support an additional speed increase when those products are introduced. It is also important to note that IEEE had a call for interest to develop chips that work on shielded systems only that would require less power. In some estimates the power consumption could be roughly 1/3 less than with the same chips that run on UTP systems. While this is not a separate project in IEEE, the development of this technology is likely to continue. Each active electronics vendor sees a significant benefit in decreasing power consumption in their equipment, which can be accomplished if they do not have to cancel noise or as much noise. This in turn would lower the cooling costs for a data centre thereby decreasing day two costs and beyond. A Quick Word About Grounding/Bonding Grounding is not as difficult as it was in years past. Most of the newer active electronics now require a ground/bond for both the power and the chassis. All grounds are terminated to a telecommunications main grounding bus bar (TMGB). This bar is then connected to a single reference ground for the facility. Problems in the past occurred when telecommunications systems were either not connected to the ground/bond points at all, or were connected incorrectly to a separate ground. The old fears are unwarranted, as standards and proper methods to create a single reference grid have existed for years. Newer connectors self terminate the shield and all ports on the patch panel are connected to the ground/bond via one connection point. Likewise, cable trays and pathways should have a ground/bond connection. Summary For anyone responsible for selecting the right cabling infrastructure and plans to occupy the premises for at least 5 years, the analysis shows that higher performing cabling systems capable of 10 gigabit transmissions or greater are the most economical solutions to provide a solid return on investment. One should consider not only the initial costs, but ensuing follow on costs as well. This includes labour to remediate a system, install increased or supplemental pathways, etc. Remember, that cabling should represent only 5-7% of the overall network investment. It is expected to outlive most network components and is the most difficult and potentially costly component of a network to replace. There is nothing worse than installing a cabling system with a shortened lifespan that will need to be replaced sooner than expected. This significantly increases the annualized cost of the system year over year as well as the overall total cost of ownership. Shielded systems are regaining popularity, and with pathway and space considerations prove to be the most cost effective system for 10G and beyond. The reality is that when you look at applications and operating system progression over the years the old adage of under utilized networks and no applications really requiring bandwidth are rapidly becoming a thing of the past. With more bandwidth available, more processing power, grid computing and other killer applications on the horizon, what we will be using day to day in 10 years will amaze! Just think 10-15 years ago and many companies were still using DOS. Where will you be in the next 10-15 years? If you rely on your data today, tomorrow it will be even more critical and in 10 years….?
|




