MAKING THE RIGHT CONNECTIONS
2014
ISSUE 014
A special supplement with
SPONSORED BY
Choosing the right cabling for your data centre
CABLINGPLANNER MAKING THE RIGHT CONNECTIONS
2013
ISSUE 013
Chairman Dominic De Sousa CEO Nadeem Hood COO Georgina O’Hara Publishing Director Rajashree Rammohan raj.ram@cpimediagroup.com +971 4 440 9131 Editorial
Ready for 40G and 100G? Upgrading you network from 10G Ethernet to 40G and 100G is not as easy as swapping out switches and line cards. Several factors have to be weighed, such as synchronising switch clocks for the higherspeeds, especially among multivendor equipment; ensuring latency remains at acceptable levels; keeping the network design and architecture optimal for 40/100G; and making sure the existing cabling infrastructure can accommodate the 4x to 10x increase in bandwidth. One of the caveats that users should be aware of as they migrate from 10G to 40/100G Ethernet is the need to ensure precise clocking synchronisation between systems -especially between equipment from different vendors. Imprecise clocking between systems at 40/100G -- even at 10G -- can increase latency and packet loss. The latency issue is a bigger problem than most people anticipate, industry experts say. At 10G, especially at high densities, just the smallest difference in the clocks between ports can cause high latency and packet loss. At 40G, it’s an order of magnitude more important than it is for 10G. This is a critical requirement in data centres today because a lot of the newer innovations are meant to address lower latencies. Another challenge is readying the cabling infrastructure for 40/100G, experts say. Ensuring the appropriate grade and length of fibre is essential to smooth, seamless operation. This is a big consideration for users because it could mean re-wiring a significant portion of their physical plant, if not all of it. That could be an expensive and disruptive undertaking. At the physical layer, 40G Ethernet is essentially 4x10G “lanes.” But 100G Ethernet is 4x25G lanes, which will be disruptive to the 10G and 40G infrastructure. Moving bits four to 10 times faster error free is a challenge in and of itself. Making sure the higher level systems -- routers and switches -- deliver services and non-disruptive service quality at those speeds is equally as challenging, if not more so. Each device has to do this at one-fourth or one-tenth the time it does at 10G. For a router, it means performing all of the packet inspection, queuing, lookups, filtering, policing, prioritization, table updating and logging while meeting SLAs by not dropping or reordering packets, or increasing latency or jitter. Upgrading your network to 40G and 100G has its own set of challenges, including multivendor interoperability, but what is most important is ensuring your network cabling is ready for this inevitable growth.
Group Editor Jeevan Thankappan jeevan.thankappan@cpimediagroup.com +971 4 440 9131 Editor Annie Bricker annie.bricker@cpimediagroup.com +971 4 440 9116 Online Editor James Dartnell james.dartnell@cpimediagroup.com +971 4 440 9140 Contributors Randy Bean Mary Brandel ADVERTISING Senior Sales Manager Michal Zylinski michal.zylinski@cpimediagroup.com +971 4 440 9119 Circulation Circulation Manager Rajeesh M rajeesh.nair@cpimediagroup.com +971 4 440 9142 Production and Design Production Manager James P Tharian james.tharian@cpimediagroup.com +971 4 440 9136 Designer Analou Balbero analou.balbero@cpimediagroup.com +971 4 440 9132 DIGITAL SERVICES Digital Services Manager Tristan Troy P Maagma Web Developer Jefferson de Joya Photographer and Social Media Co-ordinator Jay Colina webmaster@cpimediagroup.com +971 4 440 9100
Published by
Registered at IMPZ PO Box 13700 Dubai, UAE Tel: +971 4 375 1500 Fax: +971 4 447 2409 Printed by Al Ghurair Printing & Publishing Regional partner of
© Copyright 2014 CPI All rights reserved While the publishers have made every effort to ensure the accuracy of all information in this magazine, they will not be held responsible for any errors therein.
F E AT U R E Data Centre
Cabling the data centre
Known as the backbone of the business, the physical infrastructure used in data centers everywhere is undergoing a revolution with the move to highbandwidth technologies. Here is what you need to know before wiring up your data centre.
4 CABLING PLANNER
2014
M
ost technology leaders would say that the data centre is the heart of an enterprise. Yet, surprisingly, it is seldom featured prominently on the radar screen of both IT professionals and business leaders. Despite the fact that it’s important for data centers to be running 24x7, no one really knows what goes on in there and what keeps them running until the inevitable happens. Hence, it is crucial for custodians of IT infrastructures to build a resilient network-critical physical infrastructure, and especially the network cabling element to ensure that their data centres are running without interruption and that all missioncritical data is preserved and available all the time. There are a number of factors that network managers have to keep in mind while designing a data centre cabling architecture. “There is no magic formula to prepare a data centre for everything the future holds, however there are a number of fundamental areas that should be considered carefully. Collaboration across all areas of the business is first and foremost. This includes identifying the key organizational stakeholders, in order to understand the business and technology drivers relevant to a company,” says Dave Hughes, Senior Technical Manager, Commscope. He adds that the second element is to partner with technology leaders who have the proven experience, knowledge and the innovation to deliver the solution. This knowledge should include active participation with the relevant standards bodies, so as to fully understand the next generation technology and advise accordingly. This strategy is the basis for a plan that will help provide a scalable, agile and sustainable IT Infrastructure. Alberto Zucchinali, Data Centre
Solutions and Services Manager from Siemon, says as with any design, cabling architecture for the data centre needs to take into account not only today’s current facility requirements, but also those anticipated in the mid to long term. “Current and future performance requirements, flexibility and scalability, operational factors (including cooling), reliability and security are all essential when planning the architecture. These factors should be carefully considered when evaluating architectures that use Top-of-Rack (ToR) switching and Any-to-All (A2A) structured cabling configurations.” A2A structured cabling uses distribution areas that provide flexible, standards-based connections between equipment. Patch panels that mirror switch and server ports connect to corresponding panels in the distribution area via permanent links. Typically located at the end or in the middle of a row of cabinets, these distribution areas create an ‘any-to-all’ configuration where any switch port can be connected to any server port using patch cords at the patch panels. Arafat Yousef, GM of Electronics & Energy Business Group, 3M Gulf, adds another perspective: “Speed, capacity, flexibility and upgradeability are the main considerations. The cabling needs to be suitable for immediate needs but also take into account future planned requirements as cabling tends to be a longer term investment than the switching/storage. The cabling needs to be capable of current data speeds and and the same time look towards at least the next upgrade of the electronics.”
Copper vs fibre With the advent of cloud computing and highly virtualized environments, this old debate has reared its head again. Will the data centre cabling slowly move towards fibre? “The never-ending need for fast
"Multimode OM3 and om4 fibre are the best current options for data centre requiring high bandwidth applications"
Alberto Zucchinali, Data Centre Solutions and Services Manager , Siemon data transmission and efficient network performance has been spurred on by demand for server virtualisation, converged data centers, and cloud computing, as well as the advent of high-bandwidth applications like video-on-demand and social media. Fibre is currently the preferred cabling media for high-bandwidth in data centres. Multimode OM3 and OM4 fiber are the best current options for data centres requiring high-bandwidth 40/100 GbE applications. The recent development of 40/100 GbE standard by the IEEE (Institute for Electronics Engineers) will help accelerate adoption of 10GbE since higher-speed 40 GbE links are needed to aggregate 10 GbE switch connections,” says Jim Duran, Product Manager, Molex. Zucchinali agrees: “Cloud computing and outsourced services will require highly virtualised environments and speeds of 40Gb/s and 100Gb/s. IEEE 802.3ba currently specifies standards for multimode and singlemode optical fibre cabling options for both 40Gb/s and 100Gb/s Ethernet. 40GBASE-SR4 over 8 fibres (4 sending and 4 receiving at 10Gb/s) and 100GBASESR10 over 20 fibres (10 sending and 10 receiving at 10Gb/s) are supported by OM3 multimode optical fibre cabling up to 100m and OM4 multimode optical fibre cabling up to 150m. 40GBASE-SL4 and 100GBASELR4 that operate over 2 fibres are supported by singlemode optical fibre cabling up to 10km. 100GBASE-ER4, also operating over 2 fibres, is supported by singlemode optical fibre cabling up to 40km (a correlating 40Gb/s long-haul singlemode application is 2014 CABLING PLANNER 5
F E AT U R E data centre
currently under development).” OM4 delivers a link length advantage over OM3 cabling and is recommended by industry standards for deployment in new data centres. While 150m links in the data centre may be fairly rare today, the additional length can provide flexibility to accommodate future applications such as the 8 fibre 100Gb/s application (100GBASESR4) that is currently under development by the IEEE 802.3bm 40 Gb/s and 100 Gb/s Fiber Optic Task Force. This applications will operate similar to 40BASE-SR4 with 4 fibres sending and 4 fibres receiving at 25Gb/s. 100GBASE-SR4 is targeted for operation over OM4 optical fibre cabling up to 100m (the current OM3 length target in only 70m). In addition, OM4 optical fibre cabling can more easily accommodate expansions and design modifications. However, copper still has its place in the data centres, according to some vendors. “Numbers still show that the number of 10Gig ports in 2014 is evenly divided between copper and fibre. In the near future as well, copper 10Gig ports will still be considered because of price. This is why R&M continues to maintain a very good position in the copper market. We have long maintained strong focus on both fiber and copper products. While we are now a leading provider in the FTTx market and have been involved in a number of large scale projects with Telcos (ILECs & CLECs), City Development, Utilities, Universities, and Metro/Railways across the GCC, we still see a number of customer demanding
David Hughes, Senior Technical Manager – MEA, Commscope
6 CABLING PLANNER
2014
copper based solutions,” says Jean-Pierre Labry, Executive VP of R&M Middle East, Turkey and Africa. Hughes from Commscope adds that a final point to note is the pending Cat 8 (40GBase-T) copper standards. These are currently being compiled by the IEEE, TIA & ISO, with the intent of providing a 40Gb/s solution, over a standard based RJ45 infrastructure. This is a costeffective solution for data centre (EoR/ MoR) applications, adding simplicity and flexibility to the design. Now, that brings up the question – what will be the impact high-speed Ethernet technologies such as 40/100 Gbps on the structured cabling market? “In highly virtualised environments, the network connections need to be fast and reliable. The current 40/100GbE standards have been out for some time, but the 100GBASE-SR4 standard using 8 fibres will decrease strand counts within a multi-fibre channel to support that application. For the 40GbE available today, new products have been introduced both for end-toend 40GbE channels and channels where 40GbE is used at one end and then broken into four separate 10GbE channels at the other. The breakout happens using a hybrid MPO/MTP assembly. When moving from traditional 2-fibre 10GbE to 8-fibre 40GbE parallel optics, careful attention must be paid to polarity to ensure error free implementation,” says Zucchilani. Hughes adds that cabling technology has to evolve to keep pace, so partnering with industry leading experts, who actively participate with the standards bodies is fundamental. “They are best placed to advise on what is coming and more importantly how it will look from an infrastructure standpoint.” The IEEE 802.3ba and IEEE 802.3bm standards groups have and continue to evolve the fibre performance criteria based on a parallel optics approach for fiber connectivity, with the manufacturers working on solutions that will simplify the architecture for current needs and future upgradeability. IT managers, says Duran, can ensure the reliability and functionality of their computer networks by adhering to IEEE 802.3, TIA-568 or ISO 11801 industry standards. Industry standards were specifically developed to keep pace with
Arafat Yousef, GM of Electronics & Energy Business Group, 3M Gulf
the evolving and never-ending bandwidth requirements of high speed network applications and provide the foundation for a coherent upgrade strategy.
Pre-terminated cabling systems With companies looking to deploy the next generation of technology, there is no doubt that in the data centre, preterminated solutions offer the end-user a solid foundation to plan, deploy and upgrade, if designed correctly. “Pre-term solutions are ideal for data centers because links are factory tested and are made to order at specific lengths. Pre-terminated solutions ensure transmission performance and reduce field installation cost,” says Duran. “Factory-terminated and tested connections guarantee maximum channel throughput without the performance variability of field terminations. In the case of reputable manufacturers, pre-terminated fibre links are constructed of high-quality fibre optic cable and components that offer an immediate payback with guaranteed maximum channel throughput for 10Gb/s as well as support for future 40 and 100Gb/s applications, providing a ready-made upward migration path. The system’s flexible design provides scalability and allows additional links to be easily added as connectivity needs expand,” says Zucchilani. For the data centre or other highdensity computing environments, there are a range of both copper and fibre pre-
terminated trunk cable assemblies that are easily installed and offer a cost-effective alternative to individual field-terminated channels. To assess the money-saving validity of this approach for any given project, multiply the number of cable runs required by the estimated time to pull, terminate and test each cable for field termination. Then multiply the same number of cable runs by an appropriate estimated time to route and simply plug in each pre-terminated trunk. Note that pre-terminated trunk assemblies can be installed in as little as 25 per cent of the time taken for standard field termination. Compare the two time estimates and then apply the hourly fee provided by the contractor. Naturally the price of the products must also be included in the comparison, so consider the cable and connectivity required to emulate the trunk assembly that you’re evaluating. There is a role for both pre-terminated and field-terminated cabling in data centres. Where lengths are short and speeds are high (particularly for 40 Gbps and above on fibre) pre-terminated cabling is the obvious choice as it is quick and easy to install and test. This brings benefits in terms of simplicity of installation in the data centre and lack of waste and debris on site, but it requires considerable planning which may lead to longer delivery times and increased cost. But there can be many long cabling runs in large data centres where the speeds may not be so high and the use of
Jim Duran, Product Manager, Molex
pre-terminated lengths more awkward. These are still ideal for on-site termination to customise the cable length to the situation,” says Yousef.
Industry standards No matter what your technology choice is, it’s critical to follow industry standards while cabling your data centre. “In simple terms, industry standards set the minimum requirements when deploying a cabling infrastructure and/or deploying a data centre. Compliance is a fundamental part of any design and maintaining the business continuity of an organisation,” says Hughes. In essence the role of standards is much broader and involves the collaboration of a wide variety of organizations and industry experts. From protocols and interoperability through to cable media, standards play a leading role in defining how the next generation technology will perform and how it will look. Manufacturers look at how to meet and exceed these requirements from a cabling standpoint, offering enhanced channel flexibility and performance and greater margins. Indeed the leading manufacturers will actively participate in standards development from the outset. Hughes adds that ISO and TIA are arguably the most recognised infrastructure standards within the Middle East and for the most part are well aligned. However, when looking closely at the suitability of any high performance solution, it should be evaluated across the whole standards based cabling channel and not purely at the component level. Component-only compliance does not necessarily guarantee channel performance. “From a technical perspective, standards ensure interoperability between cabling system components and network equipment vendors. This is important for avoiding being bound to a single proprietary system and one vendor, which often leads to a lack of flexibility and support in the future,” says Zucchinali. When we think about interoperability there are two considerations: electrical and mechanical. It’s clear that we need network equipment to communicate, and the standards work by specifying that the various pieces of equipment must talk to one another. ISO/IEC and TIA cabling is specified in terms of electrical performance
Jean-Pierre Labry, Executive Vice-President of R&M Middle East, Turkey and Africa when connectors are mated, ensuring the integrity of the communication link between devices, even if cabling system components are mixed. Beyond pure signal transmission, the IEEE 802.3 Ethernet Working Group publishes application requirements to ensure that devices also talk the same language, he adds. Mechanical interoperability is also specified by the standards, which demand that manufacturers create products having geometry within a set range of tolerance. This ensures that systems have the same footprint for consistent and reliable mechanical connections that are capable of supporting the electrical interoperability. To date, published standards have always specified backward compatibility to ensure that higher performing systems support applications designed to run on lower performing cabling. This premise allows end users to specify systems with performance capability beyond their current needs, providing peace of mind that their cabling infrastructure will support current data transmission needs, as well as future applications requiring higher throughput. Whilst the standards bodies specify and recommend, they don’t have any power to enforce. To ensure that a system meets all parameters of a standard, end users must therefore take responsibility for verifying compliance. Independent test laboratories are an effective verifying authority, but the end user should still evaluate their reports or commission appropriate investigation work. 2014 CABLING PLANNER 7
F E AT U R E cat 8
Cat in the offing
A sneak preview of parameters of forthcoming twisted-pair cable standard for 40G
C
ategory 8, the next-generation twisted-pair cabling specification, is still in development, but the outlook looks quite positive that it will be specified to 2GHz, four times today’s bandwidth of 500MHz, promising a new copper speedway for data centres in the not-too-distant future. “What Category 8 copper cabling will do for the data center is let them transport data four times faster on essentially the same type of cable they now use,” says Sterling Vaden, chairman of TIA 42.7, the TIA subcommittee on Copper Cabling Systems. “That makes great
8 CABLING PLANNER
2014
sense application-wise and economically, because the intended data rate being developed by IEEE is 40G, four times faster than 10GBASE-T, using the same or less power per port, which has an unofficial target of less than 2 Watts.” Vaden’s subcommittee is charged with writing the standards for the performance of copper twisted-pair cabling, TIA-568-C.2, or Balanced TwistedPair Telecommunications and Components Standards. Although the standard will not be finalised for several months, Vaden says it is important to data centre managers and
suppliers to have a good sense of what Category 8 will look like and how it will affect the data centre infrastructure. Overcoming physical limitations Most likely composed of four shielded copper twisted pairs, with the similar overall diametre as Category 6A and Category 7Acables, the Category 8 cabling system under development also will use the popular modular RJ-45 style of connectors. The intention of the Category 8 standard is to be fully backwardcompatible with the previous Category 6A and lower standards.
In order to support the 40G data transmission, Vaden anticipates that the overall length of the Category 8 cabling channel will be limited to 30 meters. This has been determined
as an cccaobjective of the IEEE Next Generation BASE-T Study Group. “That is the direction the IEEE is going in terms of supporting the 40G data rate over twisted-pair cabling,” Vaden says. “We’ve performed cabling surveys in order to assess what the reach needs are within the typical data center, and the surveys indicate that the majority of the applications within the data center can be serviced with a 30-meter overall reach. The results of these surveys have also been reviewed and confirmed in the IEEE Study Group.” The biggest question asked by data center users about the new standard, according to Vaden, is how are you going to go four times faster over the RJ-45 connector? “That is the thing that people have a difficult time understanding,” Vaden says. “However, according to research performed by companies cooperating in Category 8 standards development, it looks quite possible to transport data four times faster through the connector. What we have learned is that the connectors are not the primary limiters to cabling transport speed. The insertion loss of the cable is the primary limiting factor.” Vaden says the obvious way to overcome cable insertion loss is to make the cable bigger. That’s not likely to happen because there are a number of design considerations that restrict moving to larger cable conductors than those being manufactured now. Plus, in Vaden’s view, using larger gauges would make the cable too big and bulky. Therefore, the practical gauge limit for copper, twisted-pair cables has, for all intents and purposes, already been established with Category 6A and 7A cables at a maximum of 22 AWG. How will the data transport performance of Category 8 cable improve over its predecessor versions?
cat 8 cables will physically be similar to shielded cat 6a or cat 7a cables. the conductor size will not change, nor should the diametre
“The enhancement needed to improve the performance through the wider bandwidth is basically a refinement of the manufacturing processes,” Vaden says. “Without going into a discussion of proprietary techniques, manufacturing processes must be more accurately controlled in order to have consistent performance up to these higher frequencies.” Leveraging the existing infrastructure According to Vaden, Category 8 cables will physically be very similar to shielded Category 6A or Category 7A cables. The conductor size will not change, nor should the overall diametre. “Category 6A and Category 7A cables and earlier versions have been sold into the data center for a long time,” he says, “and they are very well understood. Data cabling is much less expensive than other data center equipment. Plus, the service life of copper cabling, including Category 8, is 20 years. So, it seems very logical to suggest that when data center managers install copper cabling, they should install the highest performing cabling so that when they upgrade other equipment, they can plug it in without having to redo the cabling plan.” All things considered, Vaden says he believes the intention of TIA-568-C.2-1 Balanced Twisted-Pair Telecommunications Cabling and Components Standard, Addendum 1: Specifications for 100Ω Category 8 Cabling, is to provide a familiar and economical solution that is easy to implement. One of the key factors is the ability of 40GBASE-T equipment to do auto-negotiation, or switch over to multiple data rates, based upon the equipment connected at the other end. “Auto-negotiation is a big feature of all the BASE-T applications,” Vaden says. “That feature automatically enables equipment to adjust data transport rates, and the anticipated application would be able to auto-range to 40G and slower data transmission speeds based upon what’s plugged in at the other end. That’s a pretty big feature that networking equipment suppliers want to leverage.” 2014 CABLING PLANNER 9
inter v ie w tarek helmy
Wired for performance
Tarek Helmy, Regional Director Gulf and Middle East, South & East Africa of Nexans Cabling Solutions, answers some of the key questions related to data centre cabling.
10 CABLING PLANNER
2014
W
hat are the critical considerations users need to keep in mind while designing a data centre cabling architecture? The most important considerations are related to structure, lifetime, management and total cost of ownership. Considerations on structure, often boils down to the choice of structured cabling consisting of a fixed cabling infrastructure and patch cords. Lifetime considerations are about how many generations of switches, servers and/or storage devices should be supported by a cabling infrastructure as well as how the migration paths to the next generation should look like. i.e. from 10G to 40G, and to 100G. Management is about how to optimise operational efficiency. This can be seriously improved by implementing Intelligent Infrastructure Management solution. And finally, bringing all these considerations together by assessing the lowest cost of ownership, incorporating not just capex, but also opex side of the equation.
Do you think structured cabling in data centre move towards fibre as the adoption of cloud accelerates? Typically fibre is used for switch-toswitch uplinks, whereas copper is used for switch-to-server downlinks. This is not going to change fundamentally, although with increase of bandwidth requirements towards 40GbE and even 100GbE there will be cases where more fibre will be used due to distance limitations with copper. The interesting area of considerations when using fibre for 40G/100Gbps is the choice between multi-mode and single mode. Although single mode is an easier way of cabling, it brings much higher costs of transceivers and active equipment.
What do you think will be impact of high-speed Ethernet technologies
such as 40/100Gbps on the structured cabling market? These high speed technologies require new type of cabling. For copper this means increased frequency support and distance limitation to 30m as well as improved connectivity (such as GG45 interfaces); new standards are being defined and will arrive under the name Category 8. Important to note is that the current evolution in standard development indicates, there will be different variances of Cat8, such 8.1/ channel I, 8.2/channel II and channel Y. For multimode fibre multilane technology of 8 or 20 fibres using MPO connectivity is required. These changes will need to be included when designing a data centre architecture and require a thorough knowledge of what’s been required.
length in advance before ordering it from manufacturers.
Are pre-terminated cabling systems the most ideal for data centre
What is the role of industry standards while cabling the data centre?
Tarek Helmy, Regional Director Gulf, Middle East & Africa, Nexans Cabling Solutions
data centres will require not only good standardised cabling components and systems but also automated management
environments? ‘Plug and play’ pre-terminated systems are vital for effectively meeting current and future requirements of network installations. By eliminating the need for on-site field termination, you can significantly reduce installation time, expense and hassle. Through careful planning, you can also ensure the minimum of material is used, whilst network performance and quality are of the best possible quality. Despite the higher cost of a pre-term solution compared to the traditional cabling, the customer needs to define the cable
When it comes to making applications possible, standards’ work is essential. Today and tomorrow, Data Centres will require not only good standardized cabling components and systems, but also vastly enhanced, automated management options. Standardisation is vital in rolling that out. CENELEC EN50174-1 is a good example to include functions such as specifications, securing quality, documentation and administration, and maintenance and repair. It acts as a step-by-step guide to planning and implementation of IT cabling systems. 2014 CABLING PLANNER 11
o p ini o n R&M
Beating the heat
Alfred Tharwat, Head of Training & Data Centre consultancy, R&M Middle East, Turkey & Africa, writes about how to improve data centre energy efficiency through network convergence.
O
ver the last couple years, the topic of ‘Green IT’ has regularly peaked the interest of CIOs and IT managers across the MENA region. Smarter investments, high quality products and high density solutions have all helped reduce the ecological footprint of organisations. But despite industry-wide initiatives and high operational costs, the demand for power is on the rise and nowhere more so than in the data centre. This is particularly true in the region. DatacentreDynamics in its most recent global census found that last year in the MEA region, data centre power requirements grew by 17.5% which was well above the global average of 6.8%. The results of this study implied that in markets where the data centre industry was well established, efficiency measures were in fact being implemented. Within the region however, much needs to be done if data centre efficiency and subsequently operational expenses are to be reduced. As Dr. Thomas Wellinger, Market Manager Data Centre at R&M Headquarters in Wetzikon, Switzerland put its, “Beyond cost savings too, there is an urgent need to address data centre power consumption in order to reduce the impact of IT on the environment. Solutions such as building data centres in cold areas are unfortunately not an option in the MENA. But there are still factors which can be addressed to make power consumption in data centres significantly more efficient and one of these is network convergence.”
Same efficiency - less energy
Alfred Tharwat, Head of Training & Data Centre consultancy, R&M Middle East, Turkey & Africa
12 CABLING PLANNER
2014
First, let’s specify just what we mean by ‘network convergence’. Years ago, resources in a building would be grouped and devoted to one particular function, such as telephony, internet, security,
building infrastructure or data. Today, however, we’re seeing integrated pools of computers, storage and networking resources increasingly being shared across multiple applications. This is enabled by automatically allocating resources to specific processes, on the basis of precisely pre-defined policies. ‘Converged infrastructure’ provides enormous efficiency increases, from both technical and business perspectives. It helps reduce overall power consumption, improves cooling efficiency and enables the introduction of further energysaving measures, which we will look into in more detail below. Network convergence essentially relies on the creation virtualised server, storage and networking capacity, which is shared across multiple applications. This allows companies to reduce the footprint of all elements in the data centre. Virtualisation requires less hardware to achieve the same levels of performance and manage the same workload and supports the latest generation of energy-efficient data centre equipment.
Additional benefits In addition to the energy efficiency enhancements described above, convergence can also make it easier to implement critical functions and can improve the responsiveness of the IT department. Enterprises planning to invest in converged infrastructure offerings can look forward to another benefit: cutting down the sprawl of cables in the facility also reduces the obstacles in the raised floor’s air path. Users don’t need to store as many cables in raised floors. This doesn’t only save on materials and resources but also improves air circulation. That saves energy. In addition, servers and switches need less transmission and signal processing power when links are short and signal transmission is undisturbed. They do not heat up as much, so less cooling power is required - and therefore less energy. The introduction of next-generation technologies also allows the use of thinner cabling, which means you can take the cabling out of the raised floor and place it up above, so the ventilation in the computer room functions with greater efficiency and less energy. Valuable space
enterprises planning to invest in converged infrasracture offerings can look forward to another benefit -cutting down cable sprawl in the facility and reducing obstacles in raised floor's air path can also be saved with this type of cable management. The computer room can be smaller. That too ultimately improves energy consumption. The latest generation of cabling also has an improved noise ratio, and therefore requires less power for noise cancelling. Monitoring network ports, cables, connectors and components in real time with an intelligent infrastructure management system will also pay off. If you have a full overview and total control of the physical infrastructure, you’ll automatically use it more efficiently. You only keep operating the capacities you really need. This means energy and material consumption can be further optimised. A survey by Frost & Sullivan found that as many as 40 percent of the switch ports are ‘forgotten’ in the ongoing operation of a large data network. They remain unused because the operator does not have a full and current overview of the infrastructure. Intelligent, automatic infrastructure management can substantially improve efficiency and thereby lower operating costs.
Introducing EEE and PoE+2 An important feature of convergence is the fact that it enables the introduction of Energy Efficient Ethernet, according to the IEEE 802.3az standard. When a link is idle, the power consumption of physical layer devices is reduced by placing part of the transmission circuit into low power mode, without impacting data transmission. An EEE-defined protocol enables Ethernet devices (in LPI mode) to keep operational parameters updated. This preserves link stability and avoids disconnections. When the link is required once again, it is simply ‘woken up’ after a predetermined delay. Power losses caused by idle circuitry are a major concern, with millions of new switches being added to the already substantial installed based each year. However, the use of Power over Ethernet, or PoE, is also facilitated with network convergence. This combines power and data transmission in a single cable, allowing for extensive use of powering devices using data cabling. The original PoE standard was introduced a decade ago and supported up to 12.95 watts, but with the introduction of PoE+ in 2009, up to 25.5 watts is supported. PoE can now power devices over long lengths of data cable.
Summary
17.5%
growth in MEA data centre power requirements
In the ‘converged’ environment, Ethernet is no longer just used for transporting data, but networks an ever-growing number of devices and allows users to make the most of system intelligence. Convergence centralises management of IT resources, consolidates systems, boosts resource utilisation rates and lowers costs. This has its repercussions on data centre design as well as a significant effect on power requirements and distribution efficiency. 2014 CABLING PLANNER 13
o p ini o n Belden
Making the right choice for your data centre
As its price and power consumption continue to drop, 10GBASE-T technology is facing off with SFP+ in 10 Gb/s rack-based data centre environments that use top of rack (ToR) switches. Paul Kish, Director of Systems and Standards at Belden reviews the pros and cons of each option so you can make an informed decision and select the solution that best fits your need.
I
n today’s data centre, more bandwidth is needed to support the use of server virtualisation where multiple virtual machines are combined on a single physical host server. To accommodate the ever-growing number of operating systems and applications residing on individual servers, virtualization requires significantly increased data transmission between the servers and switches. At the same time, the amount and types of devices residing on the network has dramatically increased the amount of data that needs to be transmitted to and from storage area networks (SANs) and Network Attached Storage (NAS). Data centre managers are looking for cost-effective methods to provide more bandwidth, while optimizing power consumption and supporting modularity and scalability. A key challenge they face is that they need to create and maintain a flexible infrastructure that can support today’s and tomorrow’s IT requirements.
Reducing power consumption
Paul Kish, Director of Systems and Standards, Belden
14 CABLING PLANNER
2014
Recent advancements have allowed switch manufacturers to significantly lower power consumption on 10GBASE-T server and switch ports. While early versions of 10GBASE-T switches required up to 12 Watts per port, switch vendors now offer a range of 1.5 to 4 Watts per port depending on distance.
For short distances, the power consumption for 10GBASE-T is approaching the small form factor pluggable (SFP+) interface that has been widely deployed for 10 gigabit ToR switches and which typically consumes less than 1 Watt per port of power. With simplified electronics, SFP+ also offers better latency—typically about 0.3 microseconds per link. 10GBASE-T latency is about 2.6 microseconds per link due to more complex encoding schemes within the equipment. When you consider an environment where data transmits through three switches, 10GBASE-T therefore introduces nearly an 8 microsecond delay, while SFP+ introduces less than 1 microsecond delay. Lower power consumption and lower latency makes SFP+ well suited for large high-speed supercomputing applications where latency is a critical factor and where high port counts can add up to significant power savings.
Cost and Interoperability While 10GBASE-T switches have been available since 2008, and as an add-on Network Intercace Card (NIC) for server platforms, 2012 saw vendors begin to ship server and storage devices with 10GBASE-T LAN on motherboard (LOM) technology. As a result, more than a half million 10GBASE-T server ports shipped in 2012. As this trend continues, the shipment of 10GBASE-T ports is projected to reach upwards of 5 million for 2013. The increase in demand is helping to drive down the cost of 10GBASE-T technology. SFP+ DAC solutions available from switch vendors are often proprietary and cost more than generic SFP+ DACs and category 6A patch cords from cabling vendors. With 10GBASE-T rapidly becoming the de facto LOM technology, the use of SFP+ can means an additional cost of adapters for the servers. In comparing one of the latest SFP+ and 10GBASE-T ToR switches, the cost of 10GBASE-T ranges from 20% to 40% less. 10GBASE-T also has the advantage of being an interoperable, standardsbased technology that uses the familiar RJ45 connector and provides backwards compatibility with legacy networks via autonegotiation. The ability to autonegotiate between 1 and 10 gigabit
10GBASE-T also has the advantage of being an interoperable, standardsbased technology that uses the familiar RJ45 connector and provides backwards compatibility with legacy networks via autonegotiation. speeds allows 10GBASE-T server upgrades to occur on an evolutionary, as-needed basis. SFP+ solutions do not support autonegotiation and are limited with little or no backwards compatibility. As a standards-based, category 6A twisted-pair technology, 10GBASE-T can offer more design flexibility using a structured cabling approach to support longer distances of up to 100 meters for End of Row (EoR) and Middle of Row (MoR) topologies, as well as shorter ToR switchserver connections using category 6A patch cords. A structured cabling approach means that Category 6A cables can be field terminated on patch panels to any length for clean, slack-free cable management. SFP+ direct attach cables (DAC) are used for short distances less than 10 meters, and they are factory terminated and must be purchased in pre-determined lengths. SFP+ active optical cables (AOC) can support longer distances up to 100 meters, but at a much higher cost and with the same limitations as SFP+ DAC for point to point connections. A comparison of the 10GBASE-T implementation using EoR topology and the SFP+ implementation using ToR topology is illustrated in Figure 1. For the 10GBASE-T implementation, Category 6A patch cords connect from the servers to a patch panel in the server cabinet and from switches to a patch panel in the switch cabinet. Category 6A cables ‚ connect between patch panels in the server cabinet to patch panels in the switch cabinet located at the end of row. This implementation is referred to as a structured cable topology and allows any switch port to be assigned to any server port. For the SFP+ implementation, SFP+ DAC cables are used for direct connections
between the servers and the ToR switch in the server cabinet. The uplinks between the ToR switch in the server cabinet and the interconnection switches in the switch cabinet are over OM3/OM4 multimode optical fibre cables and OM3/OM4 optical fibre cords. These uplinks are usually 40G Ethernet using QSFP+ plug-in transceivers in the switch ports and MPO optical fibre connections between switches.
Make an informed decision When faced with choosing between SFP+ and 10GBASE-T, carefully consider your needs. If power consumption and lower latency are critical, SFP+ might be the solution for you. If cost, flexibility and scalability are more important, you’ll likely want to consider 10GBASE-T. Regardless of which solution is for you, it is important to remember that high quality cabling from reputable manufacturers will help ensure performance and reliability—especially since ToR switchserver connections are touched frequently in moves, adds and changes. Belden offers both SFP+ Direct Attached Cables for SFP+ and category 6A patch cords for 10GBASE-T—and the experts to help you choose. Our SFP+ Direct Attached Cables feature Belden Lightskew Series cable with optimal delay skew and zinc die-cast connectors with easy pull-to-release action. Part of the Belden IBDN System 10GX, our category 6A patch cords are available with an integral boot for improved strain relief and patented Bonded-Pair technology that maintains stable performance when subjected to everyday handling stresses. Belden category 6A cords are also available as Traceable patch cords for easy tracing of connections and integration with DCIM systems. 2014 CABLING PLANNER 15
o p ini o n CommScope
Planning for the future of cabling with Cat 8
In recent months, 40GBASE-T has come several steps closer to demonstrating proven viability for data centre applications. Ciaran Forde, VP – Enterprise Sales, CommScope MEA, discusses the impact of this game-changing innovation
T
he pressure on data centre capacity has never been greater. It is now commonplace for 10 or more virtual machines to be running on the same physical server, making a high-speed uplink essential to ensure all these applications can run without excess latency or degradation of service. As more and more servers migrate to 10GbE links to support this data deluge, 40G uplink speeds are forecast to be a requirement for server interfaces by 2018. CommScope recently verified a proofof-concept solution for a viable 40GBASE-T channel by utilising prototype Category 8 RJ-45 connectors and copper twisted pair cables at an IEEE 802.3 NGBASE-T study group meeting. This proof-of-concept demonstrates that data centre operators can use Category 8 cabling to support 40 GBASE-T. One of CommScope’s key intentions in helping to develop the industry standard for 40GBASE-T was to ensure that it could be supported and rolled out costeffectively. The new standard will minimise the time it will take to develop new electronics for switches and servers that can support 40GBASE-T connectivity, by building on the work already completed to support 10GbE connections. Additionally, the standard will also support the ubiquitous RJ-45 connector.
The way forward Ciaran Forde, VP – Enterprise Sales, CommScope MEA
It has been proposed in some corners that Category 7A cabling offers futureproofed support for 40GbE speeds.
However, Category 7A is only specified to a frequency of 1GHz and PHY industry experts have shown that higher bandwidth, in the range of 1.5-2GHz, will be required to support 40GBASE-T transmission. To date, the work of standards bodies indicates that a higher bandwidth solution than Category 7A cabling is needed in support of 40GBASE-T. Additionally, Category 7A supports three different types of connectors that are not interoperable or backwards compatible with the RJ-45 connector. Their use in DCs will require the use of hybrid patch cords and equipment cords making operations more difficult, error prone and slowing down MACs (moves, adds, changes) in data centres. The IEEE has made it clear that it requires a low-bulk, high-density cabling solution for 40GBASE-T, given the limited real estate in data centres. Due to these factors, the wider cabling industry has agreed that Category 8 is the path forward for supporting 40GBASE-T. CommScope successfully demonstrated the technical feasibility of Category 8 cabling for enterprise networks. This is a step along the real path towards a viable 40GBASE-T system for data center applications. Today, 10GBASE-T continues to gain broad acceptance in the server and access layers, and CommScope anticipates bandwidth growth will necessitate a 40GBASE-T solution in the future. BASE-T technology in conjunction with RJ-45 connectivity is a widely accepted and cost-effective networking option in both data center and commercial building environments. The CommScope proofof-concept demonstrates that data centre operators will be able to extend their preferred mean of communication technology for 40G.
for wholesale upgrades in 3-5 years that may generate significant down-time. For this reason, data centre operators should be careful not to reduce their initial CapEx at the cost of greater OpEx during the total lifetime of their infrastructure. Purchasing better quality equipment can also reduce the need for on-going maintenance and technical support during a data centre’s lifetime. Similarly, media diversity can also play a crucial role in ensuring reliability in data centres.
the combination of 40gbe data rates with the optimisation possible through intelligent infrastructure is a potent one. it ensures minimal risk and maximum performance of it systems well into the future 40GBASE-T: Raising the value of intelligence As the bandwidth of each individual link in the data centre rises, so does the cost of a connection failure. The huge volumes of data that will soon be running across individual cabling will mean that the failure of even one connection can have a significant impact upon a data centre’s overall performance. Unfortunately, the rapid increase in network complexity in recent years has produced many more potential points of degradation and failure in data centres - from a patchcord being accidentally unplugged to a software false alarm halting all network traffic. It is
Evolution or extinction With the requirements placed on data centres increasing rapidly, the ability to flexibly adapt to future demands is crucial for data centre managers. Often this can be achieved by deploying higher bandwidth solutions in one part of the data centre, provided that these systems are backwards compatible with existing infrastructure. Organisations that do not future-proof their data centres today, will face the need
essential then that enterprises have the tools in place to proactively track and monitor data centre activity. Businesses are rapidly waking up to the reality that the less efficient methods used in the past for maintaining their networks and tracing faults are no longer adequate. The answer lies in “Intelligent Infrastructure solutions” – systems that provide the missing link between network management tools and the traditionally
Cat8
is the path forward for supporting 40GBASE-T
passive structured cabling infrastructures that connect network devices together. Intelligence provides better, real-time monitoring of the physical network layer, helping IT professionals and network managers to ensure the security of their network by quickly eliminating blind spots, detecting and locating network breaches, and constantly keeping track of all moves, adds or changes within the network. As the data rates of individual links in data centres increase, IT managers also need to “sweat the assets” they have, as the individual connections are becoming more valuable. To this end, Intelligence inside the data centre also improves operational efficiencies across the board, providing visibility into what resources are available, what data speeds can be supported and, if maintenance is needed, what ports are available and what systems can easily be moved. The combination of 40GbE data rates with the optimisation possible through Intelligent Infrastructure is a potent one. It will allow data centre operators and enterprise IT directors to ensure minimal risk and maximum performance from their IT systems well into the future. Towards this end, it’s crucial that planning for the future starts today. 2014 CABLING PLANNER 17
Pr o d ucts round up
Siemon debuts media patch panel
R&M launches HD SCM Street Cabinet R&M, the Swiss structured cabling specialist, has unveiled its HD SCM street cabinet. This new tray system can house up to 50 percent more fiber optic connections than a similarly sized box, and has been specifically designed for connecting, distributing and managing a large number of optic fibers in a tiny space. The HD SCM street cabinet can house as many as 1728 splice connections- a world record in this segment. Up to 120 conduits with fiber optic cables can be drawn into the distribution cabinet. These conduits are fixed in place with a single click in a newly developed strain relief device thus saving the use of cable ties and screw connections. Attention was also paid to quick assembly and all components in the distributor platform can be mounted and retrofitted without the use of special tools. The splice area can be swiveled out like a collapsible table, allowing users to work all around on the trays, splices and fiber management. R&M offers the new street cabinet in four different sizes and with variable possible uses. This approach allows a flexible adaptation to various infrastructures and cabling strategies or to special site conditions. Specific adaptations and individually preterminated SCM modules can be provided to accommodate special applications, proprietary standards and national specifications. The R&M platform supports blow-in and loose-tube cabling.
18 CABLING PLANNER
2014
Siemon has launched a new mixed media patch panel for the Middle East market. The 1U ‘Copper/Fibre Combo Panel’ allows copper outlets to be mixed with fibre ports in the same rack mount space to efficiently maximise rack space utilisation. According to Siemon the new ‘Combo’ panel offers exceptional versatility and density, making it ideal for any situation where space is limited. It is compatible with all Siemon plug and play metal (PPM) fibre modules and metal fibre adapter plates, accepting up to four fibre modules or adapter plates within its 1U height. Each copper adapter plate fits up to six copper outlets with individual port identification and accommodates both shielded and unshielded copper systems, from category 5e to Siemon’s high performance category 7A TERA and category 6A Z-MAX outlets. Constructed of lightweight high
strength steel with anodised black finish, the Combo Panel is robust and, like all Siemon products, is designed to be installer-friendly. It is ANSI/EIA 310-D compliant with an integrated grounding strip to ensure proper ground path from copper outlets to grounding point. The design includes integrated cable management features to secure cables in order to provide proper strain relief and the panel is supplied complete with tie-wraps, grounding kit, and mounting screws.
Belden improves network availability through mobile monitoring
Belden has launched HiMobile, a new app from its Hirschmann product portfolio. Running on smart phones or tablets, this app improves network availability by allowing users to access status information about switches from almost anywhere. This enables administrators to respond immediately to error messages even without a PC. To avoid typing errors on the smaller keypads used on these devices, IP addresses can also be entered by scanning a QR code affixed to the switches. A further alternative is to select network devices from lists. HiMobile is particularly suitable for applications that depend on seamless processes
and reliable data communication. Possible application scenarios include machinery for the food and beverage industry, material handling systems, printing presses, and robotics. The app can be used to improve network availability in the oil and gas sector or in power generation from wind power and solar energy just as effectively as in road and rail transport or at airports or ports. “Used together with the Industrial HiVision network management software, the HiMobile app is the perfect solution for mobile monitoring of network devices,” says Product Manager Mark Cooksley. The management software acts as the server and the app as the client. Access to the server is possible via WLAN, Ethernet or a 3G mobile connection, and can be protected by a password. The status information for network devices is read-only and cannot be updated. To make changes to the network a mobile worker needs to login to the full version of Industrial HiVision. HiMobile runs on browserenabled mobile devices under the Apple, Android or Windows Phone operating systems.
Simple. Adaptable. Manageable. quick uides for Solution g ! eployment and easy d
Simple: We are committed to making our solutions the easiest to install, configure, and integrate into either existing IT systems or data centers — or new build-outs. We ship our solution as “ready to install” as possible (e.g., tool-less rack PDU installation and standard cable management features). With our easyto-configure infrastructure,you can focus on more pressing IT concerns such as network threats.
ns Configuratio ace! sp for any IT
Adaptable: Our solutions can be adapted to fit any IT configuration at any time — from small IT to data centers! Vendor-neutral enclosures, for example, come in different depths, heights, and widths so you can deploy your IT in whatever space you have available — from small IT or non-dedicated spaces to even large data centers.
manage your Monitor and ere! rom anywh IT spaces f
Manageable: Local and remote management are simplified with “out-of-the-box” UPS outlet control, integrated monitoring of the local environment, and energy usage reporting. Manageability over the network and robust reporting capabilities help you prevent IT problems and quickly resolve them when they do occur — from anywhere! What’s more, our life cycle services ensure optimal operations.
Easy-to-deploy IT physical infrastructure Solution guides make it easy to determine what you need to solve today’s challenges. The core of our system, vendor-neutral enclosures and rack PDUs, makes deployment incredibly headache-free. Easily adjustable components, integrated baying brackets, pre-installed leveling feet, and cable management accessories with tool-less mounting facilitate simple and fast installation.
Integrated InfraStruxure™ solutions include everything for your IT physical infrastructure deployment: backup power and power distribution, cooling, enclosures, and management software. Adaptable solutions scale from the smallest IT spaces up to multi-megawatt data centers.
Business-wise, Future-driven.™
Download our Top 3 Solutions Guide for FREE and enter to WIN a Samsung Galaxy Note™ 3! Visit: www.apc.com/promo Key Code: 45610p ©2014 Schneider Electric. All Rights Reserved. Schneider Electric, APC, InfraStruxure, and Business-wise, Future-driven are trademarks owned by Schneider Electric Industries SAS or its affiliated companies. All other trademarks are the property of their respective owners. www.schneider-electric.com • 998-1155587_ME_C_Note3
CNME_UAE_Jun_45610p.indd 1
5/22/14 9:39 AM
Does your fibre system tick all the boxes?
LANmark-OF : Competitive Fibre Optic Solutions 40G
100G
• Micro-Bundle cables save up to 50% trunk space • Slimflex cords offer 7,5mm bend radius saving 30% space in patching areas • Pre-terminated assemblies reduce installation time • MPO connectivity enables cost efficient migration to 40/100G
w w w.nex ans.c om/L AN s ys te m s
LANmark-OF brings the best fibre technologies together to ensure maximum reliability and lowest operational cost.
OF brochure
Accelerate business at the speed of light
info.ncs@nexans.com
Global expert in cables and cabling systems