A Suitable Solution: Fibre to play a key role in ensuring data centre connectivity

Fibre to play a key role in ensuring data centre connectivity

The demand for data centres is fast gro­wing. Data centre capacity is expected to double by 2023 to over 1,000 MW. Domestically speaking, as per NASSCOM, the In­dian data centre market can see cumulative investments of $25 billion between 2019 and 2025. Given the growing demand the Indian government came out with the Data Centre Policy in 2020. The policy seeks to ensure sustainable and trusted data centre capacity in the country to meet the enormous demand. Fur­ther, it aims to strengthen India’s position as one of the most fav­ou­rable countries for data centres by incentivising and facilitating estab­lish­ment of state-of-the-art data centres. In addition, the policy seeks to promote do­me­stic ma­nu­facturing, including non-IT as well as IT components, to inc­rease do­m­estic value addition and reduce depende­nce on impor­ted eq­uip­ment for data centres.

A look at some of the key connectivity re­quirements of data centres and the growing data centre interconnectivity market…

Fiberisation requirements

Data centres vary widely in their design and requirement. Some of the factors influencing the design of data centres are budget availability, adherence to insurance and bu­il­ding code, availability of power, prese­nce of cooling ca­pacity, adoption of site safety protocols to safeguard against disasters such as floods, seismic sho­cks, etc., availability of space for moving equi­pment, weight bearing capacity of the site and ava­ilability of uninterrupted high speed connectivity.

Of these, ensuring uninterrupted high speed connectivity is gaining increasing fo­cus among data centre operators. To meet this connectivity requirement, deploying op­tic fibre cable is being seen as a suitable so­lution. While traditional copper cables can provide only 10 Gbps of bandwidth, fi­bre optic cables can provide over 60 Tbps. Further, optic fibre cables are more immu­ne to noise, while copper cables are susceptible to EM/RFI interference, cross-talk and voltage surges. Moreover, fibre cables are more se­cu­re than copper cables as they are nearly impossible to tap. In addition, fibre ca­b­les are li­ghtweight, have a thin dia­meter, come with str­o­ng pulling strength and weigh only 4 LBs. Copper cables, in contrast, are mu­ch heavier, have a thicker diameter, strict pulling specifications and weigh over 39 LBs. Over and above this, optic fibre cables come with a life cycle of 30-50 years and ha­ve an energy consumption rate of 2 W per user. Mean­while, copper cables come with a life cycle of five years and have an en­ergy consumption rate of over 10 W per user.

Data centre interconnecting market

Data centre interconnectivity (DCI) includes both “intra-” and “inter-” data centre connectivity. In­tra-data centre connectivity offers short reach and it is cost sensitive to connect servers and storage in this scenario. Meanwhile, inter-data centre connectivity offers long reach between data centres. The global market for DCI is projected to be around $6 billion by 2025, and 50 per cent of this market is expected to be in em­erging econo­mies, dominated by Asia. DCI sp­ending is dominated by web-scale companies (such as FAANG) in developed markets and by telcos in developing countries such as India.

Key DCI requirements

  • Bandwidth scalability: DCI demands high-bandwidth connections with wavelength ca­pacities ranging from 100G to 600G. Furth­er, it requires support for up to 64 waves per fibre @ 600G per wave or 96 waves per fibre @ 200G per wave. It also requires a rich set of client interfaces (SDH, Ethernet, fibre channel) with tunable and pluggable optics.
  • Transmission security: This is another key requirement of DCI. Data encryption is possible on the client end, for in­s­t­a­­nce, Ethernet, or at frame (OTN) layers.
  • Terabit-scale switching: This is re­quired for countrywide networks. Sca­l­able multi-terabit OTN cross-connect is re­quir­ed at metro data centres to optim­ally aggregate traffic from multiple edge data centres ranging from sub-Tbps to tens of Tbps. The system en­tails pay-as-you-grow scaling through in­novative disaggregated leaf-and-spine architecture.
  • Service agility: DCI requires software defined networks (SDNs) for service agility. SDN in the transport layer enables net­work automation, which speeds up con­figuration and provisioning of tasks; network intelligence, which enables pro­ac­tive response to various network scena­rios such as failures; and network progra­mmability, which integrates third-party apps and service requests through APIs.
  • Future-proof architecture: This is yet another requirement of DCI. In­sta­llation of software-defined hardware bri­ngs the benefits of lower total cost of own­­ership and ability to deliver custom fe­a­tures for service providers. Fur­ther, for OEMs it offers the advantage of shorter time-to-market and field upgra­des on the installed base. Also, design reuse he­lps lower research and development cost.


New data centre build-outs and intra-DC infrastructure upgrades to support rising cloud service adoption is driving high capacity DCI networks. Modern DCI networks have to satisfy multiple requirements – scalable capacities with bandwidth-reach maximisation, secure and low-latency tra­n­sport, cost-effective terabit-scale swit­ching, versatile software control and future-proof programmable hardware.

Based on a presentation by Kanwar Jit Singh, Vice-President, Tejas Networks