Network cabling to reduce risk and ensure compliance
What is the history of Data Techniques?
Nick Taylor: Data Techniques was founded in 1988. It is a privately owned business with a turnover of around £28 million, of which over 90% is derived from network-cabling-related work. Data Techniques was one of the first fibre specialists in the UK and has installed network cabling into a wide variety of environments. The Farnborough office specialises in installations for clients in the defence, telco/carrier, finance and technology sectors, particularly in office and data centre environments. Data Techniques is also currently the only UK-based Panduit Platinum partner.
Why is latency equalisation necessary for clients involved in high-frequency trading (HFT), the services of which are in co-location data centres?
This is relevant to trading venues where the goal is to equalise the total latency from 'click to close'. The part that we deal with at Data Techniques is latency caused by network cabling. Reducing latency and equalising latency are, of course, two different matters.
Network cabling is just one of the factors that contributes to latency in a network but, in some cases where long cables are required, the cabling can be the largest contributor. The most widely used method to reduce this is to redesign the network layout so that users are physically closer to the application device.
Latency equalisation is a separate issue, but very much a hot topic, as Directive 2014/65/EU - commonly referred to as MiFID II - is due to come into force on 3 January 2018. Together with Regulation (EU) No 600/2014, MiFID II provides an updated and harmonised legal framework governing the requirements applicable to investment firms, regulated markets, data-reporting service providers and third-country firms providing investment services or activities in the EU.
As an example, it requires trading venues to provide to all users that 'have subscribed to the same co-location services, access to their network under equivalent conditions'. This includes the provision of power, data access services and equal cable length.
Data Techniques and Panduit are currently working with the London Stock Exchange Group (LSEG) to deliver equalised latency over new fibre cabling for trading and carrier access. LSEG selected Panduit's HD Flex system for all connectivity including OS2 and OM4+ Signature Core, which supports shortwave division multiplexing (SWDM) circuits up to 200G.
In accordance with Article 48(8) of MiFID II, LSEG will need to publish the cable lengths. In order to provide the highest level of accuracy for length testing, Data Techniques used an optical frequency domain reflectometer (OFDR), which has a measurement resolution of <3mm. It is worth noting that delivering equal length cables alone will not deliver equal latency unless other cable properties are also factored in. From an installation perspective, one of the main decisions would be whether to manage excess cable or to spool unjacketed fibre cores inside a panel or cassette.
How do you ensure that Data Techniques always complies with the industry's best practices?
There are plenty of network cabling installation companies in the UK, but few that can map the logical to the physical. In other words, there are few that can understand a logical network design and use it as the basis on which to design the network cabling best suited to meet their current and future requirements.
Most of our customers are not experts; they rely on us to inform them of changes in industry standards and best practices, and to ensure that all regulatory requirements are being adhered to.
Data Techniques is a member of the Building Industry Consulting Service International (BICSI), the Fibreoptic Industry Association (FIA) and has representation at the British Standards Institution (BSI). It attends industry events held by suppliers and manufacturers, and meets with its customers on a regular basis to share relevant information.
What would you say are the principal IT challenges that banks now face, and how can network cabling infrastructure help to address these?
Upgrading network hardware can be a challenge but it is always easier when customers can avoid the disruption of installing new cabling. For this reason, most of our customers in the finance sector install the highest-specification cabling that is able to support as many of the different network topologies as possible. Examples of suitable fibre connectivity choices include the use of multifibre push-on (MPO) connectivity to support parallel optics (circuits using more than one pair of fibres) and wideband multimode fibre (WBMMF) to support up to 200G duplex using SWDM.
What role does network cabling play in enabling 'smart buildings' to take shape?
Smart buildings require network connectivity between all systems that require control, such as if you want to use an occupancy sensor to control the air conditioning. These types of systems must be networked and integrated to provide the required level of control. The ultimate goal of most organisations is to use Ethernet as the common network protocol between these systems, which includes the associated cabling. In most cases, these are comprised of twisted-pair copper and fibre cabling.
Most customers are currently in a transition phase where mechanical and electrical systems have an Ethernet interface but use more traditional BAS protocols like BACnet and LonWorks. Issues can arise when these older systems need replacing with Ethernet versions but the cable distances exceed the maximum permissible length.
A bank must ensure that its business-critical applications are continuously transacting and eliminate downtime. Is network cabling infrastructure the answer to eliminating downtime?
Even the best network cabling design cannot eliminate downtime but a few smart decisions related to its deployment can have a big impact.
Migrating to a new cabling system is not without risk. The simplest way to reduce this is to design a system to maximise the chances of supporting future network speeds and topologies. Another contributor to downtime is patching errors: cords that should not have been removed or patched into the wrong ports.
Regularly inspecting and cleaning fibre ports and cords prior to connection also helps. One of our clients uses a leased line between two sites to support multiple circuits using dense wavelength division multiplexing (DWDM) transmission. The link had been working without issue for a few years but after some of the transceivers were replaced, packets were being dropped. The new transceivers were checked in case they were faulty, then the fibre patch cords were replaced and the internet service provider (ISP) was contacted to test the leased line. The fault was down to contamination on one of the connectors in the patch panel, probably transferred from a dirty fibre cord; but the engineers did not have an inspection microscope, so could not locate the problem.
The methodology for remedying this problem is clear and it starts with inspection. If contamination is present, clean the connector and then inspect again to ensure that it has been removed; often a dry clean is not sufficient and a wet clean is required first. If this process is not followed, there is a risk of transferring the contamination to another connector, which can be very difficult to remove because of the pressure used to keep the connectors mated. The worst-case scenario occurs when the contamination scratches the end face of the connector it mates with, which usually entails the replacement of both. All of this takes time, but it greatly reduces risk. At a minimum, we would recommend this approach is always taken on higher risk circuits.
Any customers looking to minimise risk may wish to consider adopting an automated infrastructure management (AIM) system. AIM systems add value in two key areas: mean-time-to-repair (MTTR) and for provisioning new connectivity. In some businesses, there are multiple departments responsible for installing IT hardware and, although most companies have systems to provision rack space and sometimes power (even down to the outlet), few have systems to reserve copper or fibre ports required for connectivity. Many AIM systems, such as Panduit's SmartZone solutions, can integrate with clients' existing provisioning tools, making adoption simpler. The wrong patch could cost our clients huge sums of money, so the uplift in capital and operational costs is considered worthwhile by many in the financial sector.
Does the introduction of SWDM over multimode fibre negate the need for parallel optics or MPObased connectivity?
The short answer is no; SWDM will complement parallel circuits. A few years ago, 10G (10GBASE-SR) was the maximum speed a pair of multimode fibres could support using a VCSEL light source. Higher speeds required parallel circuits; for example, 40G (40GBASE-SR4) requires eight fibres. Another method to achieve higher speeds is to use SWDM. One of the earliest examples of a transceiver using SWDM was Cisco's 40G BiDi transceiver, which uses two wavelengths (850 and 900nm) each of which supports 20G (an increase compared with 10GBASE-SR due to the reduced drive distance). By using four wavelengths (PAM-4 encoding) instead of non-return to zero (NRZ), it is possible to provide up to 200G per pair of multimode fibres. The current roadmap shows that speeds above 200G will require a combination of SWDM and parallel circuits. Parallel circuits can also support break-out connectivity. For example, a 100G port split into 4×25G circuits.