Connectivity: From land, to sea, to space by DCD Magazine

> eBook

Connectivity: From land, to sea, to space Examining the crucial role of connectivity – no matter where we are

We’ve got you connected! We’ve We’ve got got you you connected! connected!

Would you like to install trunk cables between data halls up to 70% faster?

Then get ready for Corning EDGE Rapid Connect, Would like install trunk the latest you innovation of our award-winning data Would you like to to install trunk cables cables between data halls 70% faster? centre solutions. Thanks theto new Fast-Track MTP® between data hallstoup up to 70% faster? Connectors, thefor small-size grip enables Then get ready Corning2-inch EDGE™™pulling Rapid Connect, Thentoget Corning EDGE trunks Rapid Connect, you pullready up tofor three through the latest innovation of3,456-fibre our award-winning data a the latest innovation of our award-winning data are 4-inch Complete data centre ® centre duct. solutions. Thanks to the newinterconnects Fast-Track MTP ® centre solutions. Thanks to the new Fast-Track MTP faster than ever, reduce costs, andpulling maximise your ROI! Connectors, the small-size 2-inch grip enables Connectors, the small-size 2-inch pulling grip enables you to pull up to three 3,456-fibre trunks through a you to pull up to three 3,456-fibre trunks through a 4-inch duct. Complete data centre interconnects are 4-inch duct. Complete data centre interconnects are faster than ever, reduce costs, and maximise your ROI! faster than ever, reduce costs, and maximise your ROI! ™

Are you interconnected?

Visit www.corning.com/emea/rapid-connect to discover how Corning can help with your Are datayou centreinterconnected? interconnect solution.

Are you interconnected?

Visit www.corning.com/emea/rapid-connect Visit www.corning.com/emea/rapid-connect to discover how Corning can help with your to discover how Corning can help with your data centre interconnect solution. data centre interconnect solution.

Introduction

Contents 3. Introduction 4. Making data center CLICK TO WATCH

interconnections faster and easier 6. Network Edge connectivity: Data centers on the network Edge 8. Corning case study: Altice,

Beneath our feet, resides tens of

the connectivity we experience today

millions of miles of copper wire that

extends far beyond what we ever could

makes up our telephone network –

have imagined even a few years ago.

some of it more than 100 years old.

In this eBook we take a deep dive into

Before the internet, this was how

connectivity, highlighting the crucial

Submarine cables find

phone signals were carried cross-

role it plays in our everyday lives, even

new impetus under

country, connecting us like never

when we can’t see it.

hyperscalers

before.

We examine how today’s data

But today, the proliferation

centers – and those in the future –

Portugal 10. Subsea connections:

14. Interplanetary

and subsequent popularity of

can make interconnections easier,

connections:

the internet, coupled with our

and how they can hope to handle the

insatiable appetite for technology

ever-burgeoning bandwidths headed

Interplanetary Internet,

consumption and innovation, has

our way. Is Edge the answer? Has the

digital zebras and the

transformed connectivity into a

centralized data center as we know it

disconnected Edge

veritable bandwidth behemoth,

had its day?

forcing companies to find a new way

Taking inspiration from

of carrying the deluge of data we’re so

hyperscalers the likes of Google,

relentlessly generating.

Amazon, Facebook and Microsoft, we

Enter, fiber optic cabling, which, apart from forming the backbone of our internet structure today, is far

look at the burgeoning technologies worth investing in. We take you from land, to sea, to

more capable than copper ever was

space and back again to discover how

when it comes to transporting data

the way we connect and operate data

at the speeds to which we’ve grown

centers is changing, whilst shining

accustomed.

a spotlight on the importance of

In an increasingly digitized world,

sustainable connectivity.

3 | DCD eBook • datacenterdynamics.com

18. Corning case study: Telehouse, London 20. Facility connections: Next gen data center connectivity: 400G and beyond 22. Broadcast: Watch on demand 23. Further reading

DCD eBook | Connectivity

Making data center interconnections faster and easier Connecting hyperscale data center campus buildings together with high fiber count cabling is costly and time consuming, but there is technology that can slash installation times and costs Building out data center campuses can take more time and money than anticipated, so in the current environment, with demand for data center services strong and only set to get stronger, while staff shortages beset every sector of industry, builders are looking for any way they can to speed-up construction, while sticking as close as possible to Corning’s Edge Rapid Connect fiber

budgets. One of the thorniest challenges for hyperscale data center campus builders is installing the communications cabling quickly, efficiently and correctly, ensuring each of the data center buildings

for pulling through unconnectorized

pulling grip to protect the connectors

is fully connected with as much

cables, but for very high fiber count

and pull the cable through, says

bandwidth as possible so that they

cables, such as 1,728 fibers and more,

Scott Gregg, a marketing manager at

can operate as one, hyper-efficient,

the subsequent splicing and testing

Corning Optical Communications.

hyperscale unit.

is a painful, slow and, ultimately,

However, that’s easier said than done because the industry doesn’t necessarily make it easy, especially

“With a very high fiber count cable

costly process that can take weeks to

and traditional connectorized MPO

perform.

or LC trunk solutions, the pulling grip

Furthermore, organizations only

would be so big that it wouldn’t even

for the largest installations. For

have so much duct space they can

example, standard data center

utilize. But when you get these very

fit through the duct.” However, it’s not just about the

interconnect ducts are typically four-

high fiber count cables and you put

cable, adds Angela Lambert, data

inches wide. This is not a problem

connectors on them, you need a

center market development manager at Corning. “When you add the

“Because length of the cable is key, we do offer services to contractors to effectively get an accurate length measurement” - Angela Lambert, Corning

splicing, and the set-up that the contractors have to go through to splice these very high fiber count cables to keep them managed and organized, it’s very labor intensive and it takes a significant amount of

4 | DCD eBook • datacenterdynamics.com

time. Once you’ve pulled it to the

they’ve got a ‘good’ splice – so then

to use the new technology, and

other end you’ve got to route this

they need to go back and test it,” says

it eliminates the need for fusion

massive amount of fiber into the

Lambert. “Then, they might only

splicers, cleavers, and splice

housing, and manage the splicing,”

find out during testing – sometimes

technicians (and their training).

she says.

weeks later – that they had a bad

At the end of it all, specialist

Not only that, Corning EDGE Rapid

splice and have to go back, locate the

Connect indoor trunks can take

technicians will need to manage

right one, break the splice and then

connectivity anywhere high-data

hundreds of numbered ribbons, and

resplice it. It’s very labor intensive

rates are required.

splice the right fibers while ensuring

and you don’t know whether it’s been

that they have the alignment just

done correctly until later on.”

right – ribbon to ribbon between

A typical hyperscale deployment,

Corning believes that it can slash the time it takes to install such cables by up to 70 percent. “It basically takes

the two ends of the cables they’re

she adds, might require deploying

what used to be a six-week project

connecting together.

more than 10,000 fibers – three

deployment down to two weeks,” says

cables in total. Installing, pulling,

Lambert. “Even when you have a very

‘rotate’ the ribbon when they splice

splicing and connecting all those

skilled labor force, it’s so challenging

them together, and when they splice,

cables will take around six weeks.

to do all the splicing; it’s so time

“They have to make sure they don’t

they don’t immediately know that

The solution, she suggests, is connectorized trunk cables that don’t

consuming and very easy to misalign a splice or to splice the wrong ribbon.

require splicing, with a new design

“But all that lengthy

of 24 fiber connector, called the Fast-

troubleshooting time can simply

Track MTP® Connector, which uses

be taken away. There’s much less

a two-inch pulling grip – as opposed

testing. It streamlines the entire

to the standard 3.5-inch grip – that

process.”

enables three 3,456-fiber trunk cables

One issue that contractors are

to be pulled through a tight four-inch

left to grapple with, adds Lambert,

conduit.

is accurate measurement. The

The connector, meanwhile, can still

connectorized trunk cables are made

mate with a normal MTP connector,

to order and measurements therefore

while the grip is waterproof and

have to be accurate.

can withstand up to 600 pounds

Corning’s innovative new grip up close

“Because length of the cable is key,

(270 kilos) of pulling tension. Above

we do offer services to contractors

all, perhaps, no special installation

to effectively get an accurate length

practices or equipment is required

measurement, so that there’s a ‘comfort factor’ that they’ve ordered the right product. Despite the concerns before the project started, she adds, early adopters haven’t reported any problems. “The contractors we worked with were initially concerned about this, but it turned out to be very simple. They managed it very easily and very comfortably,” she says.

Corning's innovative pulling grip can pull the fattest fiber along the slimmest of ducts

5 | DCD eBook • datacenterdynamics.com

DCD eBook | Connectivity

Data centers on the network Edge 5G infrastructure is being built from the ground up to support larger numbers of smaller data centers at the network Edge Fifth generation cellular networks have

telecommunications backbones to

information from user devices before

been specifically designed to support

a distant facility that may adversely

sending it to a local data center where

innovative applications and use

impact performance.

much of it is cached (video content for

cases with demanding performance

As such, 5G is likely to herald the

example).

requirements in terms of speed,

arrival of larger numbers of smaller data

capacity, latency and reliability.

centers situated in closer proximity to

local (or Edge) facilities via wired

The sheer number of attached 5G

Base stations transfer data to those

where data is actually created − micro

wide area network (WAN) switches (or

devices is also expected to generate

or pod type data centers for example

wireless mesh architectures in some

huge volumes of data traffic − IDC is

as well as suitable hosting architecture

cases) with local data centers then

predicting 90ZB of data from Internet

embedded in telecommunications

connected to regional or centralized

of Things (IoT) devices alone by 2025.

transmission equipment (base

hubs via Edge, access and core

But not all of that information needs to

stations, antennae towers, micro

telecommunications backbones.

be stored and processed in centralized

sites and access hubs etc) and even

data centers which offer super scale

endpoints themselves (including PCs,

applied to massive Machine Type

hosting capacities.

smartphones, and IoT devices).

Communications (mMTC) workloads

In fact, much of it will end up being

Similar topologies can be

but alter for Ultra Reliable Low-Latency

handled in facilities located at the

Topology depends on the application

Communication (URLLC) use cases

network Edge. That way, the distance

That radical shift in strategy and

which center on Internet of Things

data has to travel across the connecting

approach, coupled with the vast

(IoT) connectivity – including smart

networks between the end user device

volumes of data anticipated, is

utility grids, autonomous vehicles,

and requisite hosting/processing

inevitably driving changes to the way

remote surgery and intelligent

infrastructure is minimized. That can

that telcos and MNOs deliver network

transportation systems for example.

help bring average latencies down to

connectivity. 5G is designed to be a

Here it is likely that no local data center

around 10 milliseconds initially, and

service driven architecture, so the exact

will be required – initial filtering,

even below the 1ms mandated by

topology which links the device to the

processing and analysis can be done at

the GSM Alliance (GSMA) in later 5G

data center wherever it may be is likely

the base station itself before subsets of

implementations.

to depend on the application itself in

the data are transferred to the regional

most cases.

hosting facility.

Having localized storage and processing resources means that

For enhanced mobile broadband

information can be filtered, analyzed

(i.e. high speed mobile Internet

RAN depends on densely packed

and fed back to end users more quickly

connectivity from user smartphones)

antennae

without having to make the round

that will mean a mix of 5G, 4G and

The 5G RAN is made up of various

trip via a series of switches, routers,

WiFi base stations forming a radio

components, including small cells,

base stations, points of presence and

access network (RAN) which aggregate

transmission masts and towers, base

Gartner predicts that by 2025, Edge computing will account for 75 percent of enterprise-generated data

stations and even home hubs that

aggregate residential traffic before transmitting it to the nearest base station.

6 | DCD eBook • datacenterdynamics.com

Small cells with ranges of around

10m to a few hundred meters are crucial to delivering high bandwidth millimeter wave (mmWave) networks needed for eMBB connectivity. Because the frequency range in the 28-40GHz

5G networks will bring significant changes to the way that data centers are operated and connected

waveband mmWave uses is very short,

Changes to the core network

more antennae are needed to provide a

The RAN is not the only place where

quickly scale up bandwidth on demand

continuous connection which means

fundamental changes to network

to support those cloud services

densely packed clusters will be used

configuration are being made to

whilst being sufficiently flexible to

to serve large numbers of users at any

better support new Edge orientated

support new use cases and go to

one time.

5G architectures. A revamp of non-

market strategies through dynamic

radio 5G network segments that

provisioning and configuration

multiple input, multiple output (MIMO)

connect devices to data centers is also

capabilities.

antennae to send and receive larger

underway, including fronthaul and

volumes of data simultaneously, with

backhaul nodes that aggregate traffic,

slicing which can divide single network

massive MIMO again providing a

the core network, mobile Edge, and

connections into multiple and distinct

greater density in heavily congested

access networks.

virtual connections and assign them

Elsewhere 5G macro cells will use

areas such as cities and transport hubs.

Telcos and MNOs have spent the

That architecture is designed to

It is designed to support network

to different classes of 5G devices,

In most cases, the 5G RAN will

last few years gradually introducing

applications and data sets for example,

transmit data to local servers which

software defined network (SDN) and

with performance metrics for each

aggregate user traffic into core

network function virtualization (NFV)

(speed, capacity and reliability for

telecommunications networks over

technologies into their architecture,

example) governed by a specific service

some form of wired fiber optic WAN

primarily intending to reduce costs,

level agreement (SLA).

connection, though in some cases

speed up provisioning and simplify

ultra-high capacity 5G trunks offering

network management whilst

No one size fits all

wireless bandwidth of up to 10Gbps will

supporting greater computability and

Gartner predicts that by 2025, Edge

be used to backhaul RAN traffic.

collaboration between different telco

computing will account for 75 percent

platforms.

of enterprise-generated data for

Where network sharing with 4G infrastructure is needed or

Key changes to the core network

example. Yet enterprise workloads will

implemented, the local server will also

include redesigned signaling

still represent just over a fraction of the

aggregate data from 4G macro sites,

mechanisms and distributed servers

total – set to exceed 175 zettabytes by

as well as WiFi access points where

which again help shorten the

2025, up from 33ZB in 2019.

coverage is provided (in city centers for

transmission path between devices and

example).

the data center to lower latency and

percent) of the data will still be stored

improve application performance.

in public cloud environments, often

So called “cloud native” networks

Not only that but almost half (49

super scale centralized facilities,

specifically designed to support public,

while the amount of data stored in

private and hybrid cloud workloads

the core will more than double the

hosted in super-scale data centers are

volume stored in endpoints despite

also being implemented, primarily

the proliferation of smartphones with

designed to increase the speed of

large local flash storage capacities for

software distribution and upgrades

example (up to 256GB per device in

and speed up automated workload

some cases).

provisioning. The 3GPP has also standardized

5G networks will bring significant changes to the way that data centers

5G core network functions which

are operated and connected, but the

will be cloud native and container

number of attached devices they

based, underpinned by new security

will link and the vast amounts of

frameworks and quality of service (QoS)

information they create mean that

models for the infrastructure- (IaaS),

multiple approaches to data and

platform- (PaaS), software- (SaaS) and

application hosting will continue to

other “as a service” applications hosted

exist side by side for a long time to

in service provider data centers.

come.

7 | DCD eBook • datacenterdynamics.com

DCD eBook | Connectivity

Corning case study 1: Altice, Portugal Future-ready telecoms and data services BACKGROUND

other Portuguese-speaking African

Altice Portugal is a leading provider

countries.

of telecommunications, internet and

As well as replacing decommissioned

4-rack unit housing. In 2017, Altice launched a sample migration project with a tight deadline,

data services. To sustain expansion

sites, Altice wanted the new facility to

with 1,400 servers, 1.1 PB of storage,

and accommodate new and emerging

address challenges such as increased

and a heavy virtualization task force.

technologies, Altice commissioned the

cable density, which created hotspots

Based on 100G QSFP transceivers,

construction of a new, state-of-the-art

inside server racks and made it

replacing the typical aggregation of

data center in Covilhã.

difficult to operate and maintain the

10G SFP+ based links, the infrastructure

infrastructure. The company also

was changed from a duplex system

THE REQUIREMENTS

wanted to increase speed, power

to a 12-fiber based structured cabling

Based in Lisbon, Portugal and with

and capacity to withstand the future

system.

main hubs operating in Covilhã,

pressure on servers and network

Lisbon, and Porto, Altice Portugal

capabilities.

is the country’s largest provider of

“Corning’s EDGE™ solution was a solid foundation to prepare us for the future and for upgrades with ease.

telecommunications services. Due to

THE SOLUTION

The modular solution did not need to

the company’s increasing growth and

Altice selected Corning’s EDGE™

replace the complete infrastructure,

the evolution of service requirements,

product line, which provides industry-

because the cabling could partly

many of Altice’s data centers were

leading optical fiber cabling to deliver

be reused, decreasing provisioning

reaching the end of their life spans.

density, speed, and simplicity for the

times and supporting our efforts in

In 2011, Altice began work on the

Covilhã data center, as well as a clear

sustainability,” says Jorge Cavaleiro,

migration path for higher speeds.

data center manager, Altice Portugal.

Covilhã data center. The first phase of the center opened in 2013, with six 500

The preterminated EDGE cabling

Utilizing Corning’s EDGE cables

m2 IT rooms and an additional 9,000

was faster to deploy than traditional

allowed Altice to maximize the energy

m2 of white space for development. It

solutions, and the smaller outer

density of racks, improving efficiency.

needed infrastructure to power data

diameter enables increased and

With the flexibility and durability of the

intensive platforms like SAPO, a search

optimized system density and airflow

Corning® ClearCurve® fiber ensuring

engine and media hub popular in

inside each cabinet, and connectivity

signal stability in tight spaces, the

Portugal that continues to expand into

of up to 3456 fibers (MTP®/MPO) in a

data center can also benefit from a significantly reduced risk of network

“We started deploying Corning’s EDGE8 solution, as it is helping on port replication and breaking out QSFP ports on the equipment to SFP ports on the servers. It is easy to install and maintain, and easy to troubleshoot if anything goes wrong” - Jorge Cavaleiro, data center manager, Altice

downtime. Altice’s data center is a Tier IIIcertified facility from Uptime, and is certified LEED Gold, while its offices are certified LEED Platinum for their environmental efficiency by USGBC. By using existing EDGE cabling, Altice could make connections and begin migration as soon as possible. European Technologies Communications Portugal (ETCP), a

8 | DCD eBook • datacenterdynamics.com

trusted global logistics partner for Altice and Corning, played a huge role in facilitating the delivery and committed to the tight deadlines, providing service within three weeks of the migration beginning. With the solid foundation built using EDGE, Altice began in 2019 testing EDGE8® in the storage area network (SAN). It provides a best-in-class highdensity solution, allowing 100 percent

“The future is bright, and we have an infrastructure that is futureproof to address the challenges that we have because of Internet of Things, Artificial Intelligence, and Big Data. We are ready to start to deploy 400G infrastructures if needed. So, we are very confident that we are prepared” - Miguel Covas, head of infrastructure, Altice

fiber utilization, reduced patch cord complexity, and a future-ready solution to go beyond 100G, 200G, and 400G,

Altice wants to work with Corning to

without conversion modules.

deploy the remaining infrastructure and expand data housing and

CONCLUSION

processing capabilities to unused

In the future, Altice is looking to

space at the Covilhã data center. The

expand the infrastructure that supports

relationship with Corning represents

these data-intensive and software-

an opportunity to test and implement

defined networking solutions on

bespoke solutions that will make this

services like SAPO and beyond.

possible.

9 | DCD eBook • datacenterdynamics.com

DCD eBook | Connectivity

Submarine cables find new impetus under hyperscalers The tides of subsea cable trends are changing rapidly amid new investments and development The first submarine cables – built

the likes of Facebook, Google, Microsoft,

new projects; the four companies have

to send telegram signals – were laid

and Amazon are investing hundreds of

invested in more than 30 cable systems

in the mid-1800s and helped create

millions of dollars in a multitude of cable

over the last ten years.

the foundations for today’s Internet

projects across the world and changing

infrastructure.

the industry in the process.

The first submarine telephone cable,

“Having accounted for less than 10

“20 years ago, they were all consortium cables; all the traditional telcos all chipped in and created a

TAT-1, was laid between 1955 and 1956.

percent of total usage prior to 2012,

consortium group, and they funded

A joint project, the cable was paid for

Google, Facebook, Microsoft, and

these projects,” explains Gil Santaliz,

by a consortium of the UK Post Office

Amazon are currently invested in over

founder and CEO of New Jersey cable

(which then housed BT), the American

40 submarine cables around the world,

landing station NJFX. “Today, the folks

Telephone and Telegraph company

supporting somewhere in the region

that are supporting these projects being

(now AT&T), and the Canadian Overseas

of 250Tbps each,” says Brian Lavallee,

built are smaller groups, and they’re the

Telecommunications Corporation.

Ciena submarine networking solutions

companies that are social media and

The first fiber optic cable came a little

expert. “Apple, Dropbox, Netflix, Alibaba,

cloud companies, because they realized

over 30 years later, and again was laid

Akamai, and Limelight represent much

they needed the owner economics, they

in partnership between AT&T, France

of the remainder.”

needed to control the architecture.”

Hyperscalers take to the seas

investors in subsea cables is search

cables in operation or development.

Where once telecoms and

engine giant Google. According to

And while telecoms companies are still

communications companies ruled the

SubmarineNetworks.com, Google has

very much part of the picture, trends are

seas and shared ownership of subsea

invested in at least 19 submarine cables

increasingly shifting away from telco

fiber, the hyperscalers are increasingly

since 2010, including six private cables.

consortiums and more towards cable

investing in submarine cables both as

projects led by Over-The-Top (OTT)

part of traditional consortiums and as

investment was the Unity cable

hyperscalers.

private projects. Where they were once

around 2010. The search giant joined

major customers of subsea capacity,

a consortium of Bharti Airtel, Global

they are now peers or even leaders on

Transit, KDDI Corp., Pacnet (now

Télécom, and British Telecom.

One of the biggest hyperscale

Today there are almost 500 subsea

Just as how they’re making some of the largest investments in data centers,

Where once telecoms and communications companies ruled the seas and shared ownership of subsea fiber, the hyperscalers are increasingly investing in submarine cables both as part of traditional consortiums and as private projects

Google’s first submarine cable

Telstra), and SingTel to build a cable connecting Chikura, Japan to Los Angeles, US. Today the company is making largescale investments in both private and consortium cables that are amongst the biggest in the world and has several other cables in development; the 16 fiber pair, 350Tbps Grace Hopper cable

10 | DCD eBook • datacenterdynamics.com

involved in a project that met four times a year on one project.” While the hyperscalers are happy to share capacity on cables, it’s unlikely they’ll start reselling capacity. Alan Mauldin, research director at TeleGeography, previously told Wired that they “would never do that” because they would then become a carrier and subject to being licensed as a carrier. In the same piece, Urs Hӧlzle, a senior vice president of technical infrastructure at Google’s Cloud division confirmed that Google, “doesn’t want to be a service provider as an ISP” and is going it alone because consortiums can slow the process of building new cables.

“The old adage of ‘build it and they will come’ no longer works” - Carl Grivner, CEO, Global Cloud Xchange

Hyperscalers want control of cables and capacity The drive by the hyperscalers and other OTT providers is driving up cable

will connect the US, UK, and Spain,

The company claims the 6,400km 12

capacity at a rapid pace; Telegeography.

and is due to come online in 2022. Its

fiber-pair cable is capable of 250Tbs,

com reports that the amount of

Equiano Portugal-to-South Africa cable

making it the world’s highest capacity

international capacity deployed by

is due for completion later in the year.

submarine cable. That will be surpassed

content networks rose over nine-fold,

Announced in June, Firmina will run

by the company’s 340Tbps UK-US Grace

to 962Tbps, between 2015 and 2019.

from the East Coast of the US to Las

Hopper cable, which recently landed

The research firm says that having

Toninas, Argentina, with landings in

in Cornwall and is due to come online

accounted for less than 10 percent

Praia Grande, Brazil, and Punta del Este,

soon.

of total usage prior to 2012, content

Uruguay.

Even if new cable projects are

“Systems built even 20 years ago lack

funded through consortiums, the

providers’ share of total cable capacity was over 65 percent in 2020.

adequate capacity and important new

OTT hyperscalers are happy to partner

“ISPs are building their own

technical capabilities, and can no longer

together; Facebook has partnered with

submarine cable infrastructure to

manage the job,” says Mark Sokol, senior

Google, Microsoft, and Amazon for

increase capacity, offer richer services,

director of infrastructure at Google

different cable projects.

and deliver international access to their

Cloud. “Updated technology is required

Google and Amazon have both

coveted content,” says Ciena’s Lavelle.

to address this continued growth in

invested in the US-Denmark HAVFRUE

“This is certainly having a positive effect

demand for capacity, driven by cloud

cable, while Microsoft and Amazon both

on the wider submarine cable industry,

computing, future Internet advances,

own fiber pairs on the Spain-US MAREA

as it results in more competition

backend replication, and the many

cable. Facebook is part of both projects,

and route diversity for increased

breakthrough innovations that lie ahead.

as well as a number of others including

availability, therefore accelerating

“We are building out our network to

the Echo and Bifrost systems to connect

access to increasingly reliable global

provide the best possible experience

the West Coast US to Singapore and

connectivity.”

to our end users and customers. By

Indonesia.

designing and funding subsea cables we

“These companies have dedicated

As well as new cables, existing cables are being continually upgraded with

are able to plan effectively for the future

teams and in-house talent that know

new technology. Following a trial earlier

capacity needs of our customers and

how to manage a project,” says Santaliz.

this year, the trans-Atlantic MAREA

users around the world, and add a layer

“They know how to partner with others

cable – originally designed with a

of security beyond what’s available over

for like-minded opportunities, and

capacity of 160Tbps – saw its potential

the public Internet.”

they’re working on multiple projects at

capacity increase from an already

The search giant’s Virginia-to-France

the same time. It’s not the way it was 20

upgraded 200Tbps to 224Tbps.

Dunant cable went live earlier this year.

years ago when you have 22 companies

Santaliz says much of the rapid

11 | DCD eBook • datacenterdynamics.com

DCD eBook | Connectivity

progress being delivered in cable technology is being driven by the hyperscalers demanding more of their suppliers. “Google and Facebook have really been the pioneers and pushing these companies like SubCom and Alcatel to up their game,” he explains. “They’ve been asking them for better solutions, they’ve done a great job of hiring talent inside their companies to push the envelope and get the manufacturers to invent more. They’ve reinvented the industry in the last five years. “They’re no longer just sitting there waiting for a proposal, they’re driving innovation they’re driving applications are sitting side by side with the telecom operators, and in some cases they’re

companies follow? Santaliz says only the largest

they have an owner’s perspective now they’re not just the customer anymore.”

telling them here’s a better way of doing

enterprise customers that will take

it.”

the time to learn the subsea industry

been investing in data centers and

Santaliz says one of the reasons the

While the biggest OTT players have

and learn the architecture to do it

cables, smaller content companies

large hyperscalers are investing in

themselves on subsea cables, but notes

have been more focused on Content

cables is to ensure continuous uptime

that the banking industry is one that

Delivery Networks (CDNs) and ensuring

and connectivity; lost connectivity hurts

could be next.

their content is cached close to their

the brand.

“The banking industry has

“Whether your brand is social media

always been just a customer to

customers. But could that change in the future?

or banking you need to make sure your

telecommunications, and they’ve been

network is always available. This round

watching the OTTs go from being

mobile at Cambridge Consultants, has

of investments for subsea networks

customers to partners. Transactions are

previously discussed with DCD the

[is the likes of] Facebook, Google,

important; the idea of not being able to

possibility of large content providers

Microsoft; [but] I’m even starting to see

process a payment or finalize a trade or

like Disney jumping into Internet

US banks invest in subsea cables.”

close on a deal because your app’s not

infrastructure through High altitude

available, means they realize that ‘never

pseudo-satellites (HAPS), perhaps the

down’ and their brands are related.”

same could happen with subsea cables.

Will other OTT players want to invest in cables?

Derek Long, head of telecoms and

We may already be seeing it; Morgan

“If we will see more OTTs outside of

With the large cloud players only

Stanley has bought 49 percent of Altice’s

the ‘Big 5’ take this route is unclear, but

growing their investments in data

regional fiber network. And Santaliz

it is highly likely within the next five

center infrastructure globally –

notes that while they might have

years,” says Mattias Fridstrom, chief

Microsoft alone has said it’s on course

bought the infrastructure as a financial

evangelist at Telia Carrier.

to build 50 to 100 new data centers each

investment, it is certainly also going to

year for the foreseeable future – their

improve the firm’s digital resiliency.

demand for new cables and increased

“Morgan Stanley’s not going to have

Where do telcos fit in an OTT world? As well as driving innovation around

capacity is unlikely to be satisfied

an outage on anything related to New

the technology, the rise of OTT players

any time soon. But will other large

York Metro on that network because

investing in their own cables means the rest of the cable industry must also

“Systems built even 20 years ago lack adequate capacity and important new technical capabilities, and can no longer manage the job” - Mark Sokol, senior director of infrastructure, Google Cloud

adapt how they do business. In July, Global Cloud Xchange cancelled its Italy-to-Singapore Eagle cable because of a lack of an anchor tenant. “The old adage of ‘build it and they will come’ no longer works,” CEO Carl Grivner recently told Capacity.

12 | DCD eBook • datacenterdynamics.com

‘newer’ areas like Africa, I can still see

relationship the carriers want with the

Eagle – just no anchor tenant. It’s an

“There was nothing wrong with

Telcos being involved to secure a lower

OTT players in the future.

OTT [over-the-top] world and if you’re

cost path towards other regions for

getting a consortium of wholesale

their domestic traffic. In Europe and

their own decisions on whether to work

carriers together you need an anchor

North America, Telcos needs to focus

directly with the OTTs as partners in

tenant. I looked at the business case and

on meeting up subsea cables for further

a new cable or maintain a customer

it wasn’t working.”

transport inland on their land networks.”

relationship selling them capacity.”

For the ISPs, Ciena’s Lavalle says he

“Each cable owner needs to make

Santaliz says the carriers aren’t going

In the longer term, Michael Ourabah,

expects to see a 16 percent decline in

to be the ‘go-to guys’ when it comes to

CEO of infrastructure and connectivity

wholesale pricing year-on-year, and a

innovation, and will instead invest in

provider BSO, warns the ecosystem

28 percent decline by 2027.

“interesting architectures.” An example

around cable operations needs to grow

he gives is Telecom Italia and Omantel’s

in order to keep up with demand, lest the

addition to lower costs per bit to remain

Italy-to-India Blue-Raman cable,

hyperscalers take over.

financially viable,” he says. “Pricing is,

which bypasses the overcrowded Suez

“The supporting ecosystem of

therefore, the main threat to business.”

Canal by going overland into Israel for

repair assets like cable repair ships

a portion of the cable. Google is also

and specialist equipment vendors will

involved in the project.

not have enough capacity to support

“ISPs are looking for more capacity in

As incumbent cables are retired and hyperscalers make more direct investments in new cables, the

“They’re going to be rewarded

the growing volume of subsea cables;

traditional economics for telcos may not

because they were innovative and

that could lead to the development of

stack up as they once did. But despite

they were able to solve for that issue.

a two-tier system where high-volume

this, they still have a role to play in the

That’s where the carriers will find their

routes run by hyperscalers stockpile

industry.

place and in providing their traditional

repair assets and are well served by them,

“On the larger traffic paths (cross-

values of point-to-point solutions

but lower volume routes are not well

Atlantic and cross-Pacific), I think we

and partnering with the OTTs and

provisioned and deteriorate,” he says.

have seen the last consortiums with

partnering with the enterprise.”

“The only way to remain competitive

telcos involved,” explains Fridstrom.

Gavin Rea, CTO at Gulf Bridge

“Most new cables will be built by one

International (GBI), notes that some

OTTs and hyperscalers and look to offer

or a few owners with a fiber pair as the

cable owners have decided not to

complementary services that extend the

least denominator.

involve OTTs directly in their cables,

reach of networks to areas where cables

and much depends on the kind of

cannot yet readily reach.”

“On more un-traditional routes to

for consortiums will be to cooperate with

Subsea cables as an alternative to inter-country ground fiber While hyperscalers are investing in a number of large-scale international projects, there’s still plenty of room for carriers to invest in ‘alternative’ routes, and Santaliz suggests domestic subsea cables could be one such area of opportunity. In the US, the Confluence-1 cable will run across the US eastern seaboard from Wall Township, New Jersey to Miami, Florida via Virginia Beach, Virginia; Myrtle Beach, South Carolina; and Jacksonville Beach, Florida. The cable will run to 2,571 kilometers, providing 24 fiber pairs and offering more than 500Tbps in capacity once complete. “In the US, we never dreamed that we needed to go in the water to connect between the US, we assumed that we'd use railroads, use highways, and we would just interconnect the US with those arteries that we've always had, because it was good enough,” says Santaliz. While there have been some inter-country subsea cables in the likes of Brazil before, they are becoming more common and Santaliz says this trend is due to continue. Australian company Fiber Expressway is planning a 10,000-kilometer, 16-fiber pair carrier-neutral subsea cable that would have seven cable landing stations between Perth and Darwin in the northwest of the country before running north to Indonesia, Singapore, and Malaysia. In India, Reliance Jio’s IAX will link Mumbai in the west with Chennai in the east via Sri Lanka, with the possible addition of Vizag in the future, before landing in Thailand and Singapore. Santaliz says that amid ever-growing capacity demands, subsea cables now offer the economics and resiliency that make domestics subsea cables increasingly viable, especially in and around the US. “What we're realizing now is that it's actually more economical and it's safer to put a cable in the water to connect the US north and south. If you had brought that up 10 years ago, you'd get laughed out of the room, now the reality is the economics are better to be in the water, you're going to get a more resilient network.”

13 | DCD eBook • datacenterdynamics.com

DCD eBook | Connectivity

Interplanetary Internet, digital zebras, and the disconnected Edge Delay-tolerant networks ask you to imagine an Internet where connectivity isn’t guaranteed Were you to have traveled through

setup is not economically or physically

it is a careful hop-by-hop approach

central Kenya in the early 2000s, you

possible.

enabling asynchronous delivery.

may have come across something

In the case of ZebraNet, each equine

The plains of Africa are one such

highly unusual: a dazzle of zebras

locale, especially in 2001 when

served as a node, equipped with a

sporting strange collars.

Kenya had virtually no rural cellular

collar featuring solar panels, GPS, a

connectivity, and satellite connectivity

small CPU, flash memory, and radio

bizarre fashion show, but rather early

required bulky, power-hungry, and

connectivity.

pioneers of a technology that could one

expensive equipment.

The animals were not part of a

day span the solar system, connecting other planets to one giant network.

Zebras care not for connectivity; they

Instead of communicating with satellites or telecoms infrastructure,

don’t plan their movements around

the migratory habits of each zebra

The connected world we inhabit

where to find the best WiFi signal. And

are stored on the collar. Then, when

today is based on instant gratification.

that was a problem for an international

the animal is near another electronic

“The big issue is that the Internet

group of zoologists and technologists

equine, it shares the data. This

protocols that are on the TCP/IP

who wanted to track them.

continues until one of the zebras

stack were designed with a paradigm

Faced with a landscape devoid of

passes a mobile base station – perhaps

that ‘I can guarantee that I send the

connection, the team had to come up

attached to a Range Rover – and it

information, and that information

with a way to study, track, and collect

uploads all that it has collected.

will be received, and I will get an

data on zebras – and get that data back

acknowledgment during an amount

from the field.

of time that is quite small,’” Professor

To pull this off, the group turned

“It was one base station for about 10-12 collars,” project member and Princeton University Professor

Vasco N. G. J. Soares explained, over

to a technology first conceived in

Margaret Martonosi told DCD. “The

a choppy video call that repeatedly

the 1990s – delay or disruption-

main limit on storage capacity had

reminded us what happens when that

tolerant networking (DTN). At its core

to do with the physical design of

paradigm breaks down.

it’s the idea of ‘store and forward,’

what would fit on the circuit board

where information is passed from

and inside the collar module. Our

Transmission Control Protocol (TCP)

node to node and then stored when

early papers did some simulation-

still serves as the de facto backbone of

connectivity falls apart, before being

based estimates regarding storage

how our connected age operates. But

sent to the next link in the chain.

requirements and likely delivery rates.”

there are many places where such a

Instead of an end-to-end network,

Fifty years on from its invention, the

It’s an idea that sounds simple on the face of it, but one that requires a

In the case of ZebraNet, each equine served as a node, equipped with a collar featuring solar panels, GPS, a small CPU, flash memory, and radio connectivity

surprisingly complex and thought-out approach to DTN, especially with more ambitious deployments. “How much information you need to store depends on the application,” Soares explained. “So this means that

14 | DCD eBook • datacenterdynamics.com

Another approach, ‘PRoPHET,’ applies probabilistic routing to nodes when they move in non-random patterns

its final destination. Adding geographic routing could mean that it only sends it forward when it meets a node that is physically closer to the end state, or is heading in the right direction.

you need to study the application that

and disconnecting nodes, without

you’re going to enable using this type

sending it down dead ends or causing a

protocols that see each node send it

Then there are multiple-copy routing

of connection, and then the amount

bottleneck somewhere in the middle?

to a bunch of others. Versions of this

of storage, and also the technologies

This remains an area of vigorous

approach like the ‘epidemic protocol’

that are going to be used to exchange

debate, with multiple approaches as

would spread data across a network

information between the devices.”

to how one should operate a DTN

rapidly, but risk flooding all the nodes.

currently being pitched.

You also need to decide how to get data from A, the initial collection point,

The most basic approach is single-

“On a scenario that has infinite resources, this will be the best protocol,”

to Z, the end-user or the wider network.

copy routing protocols, where each

Soares said. “But in reality, it’s not a

How do you ensure that it travels

node carries the bundle forward to the

good choice because it will exhaust the

an efficient route between moving

next node it encounters, until it reaches

bandwidth and the storage on all the

15 | DCD eBook • datacenterdynamics.com

DCD eBook | Connectivity

nodes.” ‘Spray and Wait’ tries to build

of the globe, such spaces for DTN

on this by adding limits to control the

are shrinking. “The spread of cell

its work, but whether all of its research

flooding.

connectivity across so much of the

is out in the open is yet to be seen.

world has certainly been helpful for

However, DARPA was also instrumental

applies probabilistic routing to nodes

overall connectivity and does supplant

in funding the development of the

when they move in non-random

DTN to a degree,” Martonosi admitted.

TCP/IP-based Internet, which was

Another approach, ‘PRoPHET,’

The agency has published much of

patterns. For example, after enough

“On the other hand, the cost of

study, it would be possible to predict

cell connectivity is still high (often

general movement patterns of zebras,

prohibitively so) for many people.

and I started to work on the Internet,

and build a routing protocol based

From a cost perspective, collaborative

we published our documentation in

upon it.

dynamic DTNs and mesh networks

1974,” TCP/IP co-creator Vint Cerf

seem like a very helpful technology

told DCD. “Right in the middle of the

direction.”

Cold War, we laid out how it all works.

Each time data travels through the network, it is used to update the probabilistic routing – although this

carried out in public. “The irony is that when Bob [Kahn]

Following ZebraNet, Martonosi

“And then all of the subsequent work,

can make it more brittle to sudden,

worked on a DTN system to connect

of course, was done in the open as well.

unexpected changes.

rural parts of Nicaragua, C-Link, and

That was based on the belief that if the

SignalGuru to share vehicle data. Due

Defense Department actually wanted

Politécnico de Castelo Branco, Soares

to increasing connectivity, such efforts

to use this technology, it would need to

combined geographic routing with

“have not caught on widely,” she said.

have its allies use it as well, otherwise

Spray and Wait to form the routing

“But you can see aspects of these

For his work at the Instituto

protocol ‘GeoSpray.’ “My scenario was assuming

you wouldn’t have interoperability

techniques still around – for example,

for this command and control

the Bluetooth-based contact tracing

infrastructure.

vehicles moving and data traveling

apps for Covid-19 are not dissimilar

Then, as the technology developed,

between them, and so I would need

from some aspects of ZebraNet and

“I also came to the conclusion that the

the geographic information,” he said.

C-Link’s design.”

general public should have access to

“A single copy is the best option if

Terrestrial DTN proponents now

this,” Cerf recalled. “And so we opened

you can guarantee connection to the

primarily focus on low-power IoT

it up in 1989, and the first commercial

destination, but sometimes you have

deployments, or situations where

services started. The same argument

to use multiples to ensure that you will

networks have been impacted – such

can be made for the Bundle Protocol.”

find someone that will get it there for

as natural disasters, or battlefields.

you eventually.”

Indeed, the US Defense Advanced

Each approach, the amount of

With the DTN Bundle Protocol (published as “RFC5050”) Cerf is

Research Projects Agency (DARPA)

not content with ushering in the

storage, and how long nodes store data

is one of the largest funders of DTN,

connected planet. He eyes other worlds

before deleting it, must be crafted for

fearing that the connectivity-reliant US

entirely.

the application.

military could be easily disrupted.

In South Africa, a DTN deployment

“DTN represents a fundamental shift

“In order to effectively support manned and robotic space exploration,

was used to connect rural areas.

in networking protocols that will result

you need communications, both for

Communities used e-kiosks to send

in military networks that function

command of the spacecraft and to

emails, but the data was just stored on

reliably and securely, even in the

get the data back,” he said. “And if you

the system. When a school bus passed,

changing conditions and challenging

can’t get the data back, why the hell

it transferred the data to the bus,

environments where our troops must

are we going out there? So my view

brought it to the city, and connected

succeed now and in the future,” BBN

has always been ‘let’s build up a richer

it to the wider net. When it returned, it

Technologies CEO Tad Elmer said after

capability for communication than

brought any replies with it.

his company received $8.9m from

point-to-point radio links, and/or bent

DARPA to explore battlefield DTN.

pipe relays.’

But as we connect every inch

“That’s what’s driven me since 1998.”

“When Bob [Kahn] and I started to work on the Internet, we published our documentation in 1974 - right in the middle of the Cold War, we laid out how it all works” - TCP/IP co-creator Vint Cerf

DTN is perfect for space, where delay is inevitable. Planets, satellites, and spacecraft are far apart, always in motion, and their relative distances are constantly in flux. “When two things are far enough apart, and they are in motion, you have

16 | DCD eBook • datacenterdynamics.com

With the Artemis Moon program, the Bundle Protocol may prove crucial to connecting the far side of the moon, as well as nodes blocked from line-of-sight by craters

Cerf asked.

to aim ahead of where it is, it’s like

again, this time it will expect sufficient

87, but still blasting on,” said Cerf. “I

shooting a moving target,” Cerf said.

communication capabilities.

sent him a note saying, ‘look, here’s

“So I went to the best possible source for this question, Leonard Kleinrock at UCLA.” Kleinrock is the father of queuing theory and packet switching, and one of the key people behind ARPANET. “He’s still very, very active – he’s

“It has to arrive at the then when the

“We can smell the success of all this;

spacecraft actually gets to where the

we can see how we can make it work,”

nodes, and I’ve got a traffic matrix, and

signal is propagating.”

Cerf said. “And as we overcome various

I have this DTN environment, how do I

and sundry barriers, the biggest one

calculate the capacity of the system so

notion of ‘now’ is very broken in these

right now, in my view, is just getting

that I know I’m not gonna overwhelm

kinds of large delay environments,” he

commercial implementations in

it?”

noted, adding that the harsh conditions

place so that there are off-the-shelf

of space also meant that disruptions

implementations available to anyone

“two pages of dense math saying, ‘okay,

were possible.

who wants to design and build a

here’s how you formulate this problem,’”

spacecraft.”

Cerf laughed.

Across such vast distances, “the

What we use now to connect our few solar assets relies primarily on line-

There’s still a lot to work out when

the problem, I’ve got this collection of

Two days later, Kleinrock replied with

Kleinrock shared with DCD the

of-sight communication and a fragile

operating at astronomical distances, of

October 2020 email exchange in which

network of overstretched ground

course.

the two Internet pioneers debate what

stations.

“Because of the variable delay and

With the Bundle Protocol, Cerf and

the very large potential delay, the

Kleinrock described as an “interesting and reasonably unorthodox question.”

the InterPlanetary Internet Special

domain name system (DNS) doesn’t

“Here’s our situation,” Cerf said in the

Interest Group (IPNSIG) of the Internet

work for this kind of operation,” Cerf

email, outlining the immense difficulty

Society hope to make a larger and more

said. “So we’ve ended up with kind of a

of system design in a network where

ambitious network possible in space.

two-step resolution for identifiers. First

just the distance of Earth to Mars can

you have to figure out which planet are

vary from 34 million to 249 million

been successfully trialed by Martian

you going to and then after that you

miles. “The discrete nature of this

rovers Spirit and Opportunity, while the

can do the mapping from the identifier

problem vs continuous and statistical

International Space Station tested out

to an address at that locale, where you

seems to make it much harder.”

the Bundle Protocol in 2016. “We had

can actually send the data.

An earlier version, CFDP, has already

onboard experiments going on, and

“In the [terrestrial] Internet protocols,

Kleinrock provided some calculations, and referenced earlier work with

we were able to use the interplanetary

you do a one-step workout – you take

Mario Gerla and Luigi Fratta on a Flow

protocol to move data back and forth –

the domain name, you do a lookup in

Deviation algorithm. He told DCD, “It

commands up to the experiments, and

the DNS, you get an IP address back

suggests the algorithm could be used

data back down again,” Cerf said.

and then you open a TCP connection

where the capacities are changing,

to that target. Here, we do two steps

which means that you constantly run

Bundle Protocol may prove crucial to

before we can figure out where the

this algorithm as the capacity is either

connecting the far side of the Moon, as

actual target is.”

predictably changing or dynamically

With the Artemis Moon program, the

well as nodes blocked from line-of-

Again, as with the zebras, cars,

sight by craters.

and other DTN deployments,

“Artemis may be the critical turning

changing.” Cerf said that Kleinrock proved

understanding how much storage each

immensely helpful. “Now, I didn’t get

point for the interplanetary system,

space node should have will be crucial

the whole answer. I still don’t have the

because I believe that will end up being

to its effective operation.

whole answer,” he said. “But, I know

a requirement in order to successfully prosecute that mission.” DTN could form the backbone of

But working that out is still an open question. “If I know where the nodes are, and I know the physics, and I

I have one of the best minds in the business looking at the problem.” As with many other aspects of the

Artemis, LunaNet, and the European

know what the data rates could be, how

Interplanetary Internet, “this is not a

Space Agency’s Project Moonlight.

do I know I have a network which is

solved problem,” Cerf said.

As humanity heads into space once

capable of supporting the demand?”

“But we’re on it.”

17 | DCD eBook • datacenterdynamics.com

DCD eBook | Connectivity

Corning case study 2: Telehouse North, London Flexible and future-ready intrabuilding connectivity Telehouse North is Europe’s first

Most importantly, the new central hub

The proposal was for a future-ready,

purpose-built, carrier-neutral colocation

had to be ready for operations within

passive cabling infrastructure solution,

data center. One of four Telehouse

three months. From an infrastructure

including planning and installation.

data centers located at the Docklands

perspective, the design had to meet a

In addition, Corning provided the

campus in London, it is the primary

number of objectives:

consultation and training required

home of the London Internet Exchange

• High density: Optimize space by

to accelerate knowledge transfer and

(LINX) and one of the most connected

maximizing port capacity within a

operational readiness to meet the

data centers in Europe.

small footprint

strict deadlines for a complete turnkey

It provides end-to-end information and communications technology

• Project delivery: Complete delivery

solution.

and installation within a strict project

(ICT) solutions including managed

timeline

services, integrated communications

• Resilience: Support 100 percent

THE SOLUTION Telehouse required a resilient solution

services, virtualization services, content

diverse routing of connectivity to

with flexibility to meet future growth

management and system security

customer suites for high availability 24

and change. Key to the design was

services, as well as disaster recovery

hours a day, seven days a week

the addition of a central hub with

• Flexibility: Easy to manage and

services. THE REQUIREMENTS

connectivity to each floor and a

requirements

customer colocation suite with enough

• Scalability: Provide flexible

Telehouse’s cabling infrastructure required an update in order to maximize

intrabuilding connectivity and the

capacity in its 9,717 square metre,

capacity for future growth

highly secure colocation centre, which

100 percent diverse routing of fiber

accommodate changing business

• Cost effective: The solution and all

capacity to meet future demand. Corning was invited to tender for this project along with two other connectivity infrastructure suppliers.

consists of 32 suites over multiple

installation tasks had to be managed

Telehouse informed Corning that they

floors. A central hub room required a

within the given budget.

selected their solution because it offered

new cable management solution that would provide flexible and future-ready,

the most value while also meeting all of To meet this challenge, Corning

intrabuilding connectivity to each of

enlisted the help of Kinetic IT, one

the five floors and customer colocation

of their preferred installers, to work

suites.

closely with Telehouse on the design.

the project requirements, including: • 2.5 times increased port density, maximizing capacity and saving space • Guaranteed two-week delivery of

“Working closely with Steve Gentle and his team at Telehouse, we were able to understand how they would best utilize the infrastructure and take into account both their current and future capacity requirements.” - Tom Cella, managing director at Kinetic IT 18 | DCD eBook • datacenterdynamics.com

of cables to each suite. Completed in March 2016, within a 12-week timeframe, the installed system has the capacity to allow for expansion up to 130,000 ports with the use of additional cabinets. Steve Gentle, Telehouse senior customer installations manager, commented, “We are very pleased with the level of technical assistance and support we received from Corning and their partner Kinetic for our fibre infrastructure project at Telehouse North. “Corning offered excellent technology and product knowledge and Kinetic supplied a high level of practical design and installation expertise. Both organizations were easy to work with and surpassed our expectations – working with the combined team was a very positive experience.” and functionality without sacrificing

CONCLUSION

density. Each cassette contains fiber

The Centrix system combines extreme

guides and a splice section and can

flexibility and simplicity with the

hold 24 or 36 LC connector adapters.

ultimate in density. With superior

fiber cabling infrastructure Corning

Telehouse personnel can easily access

jumper management and an innovative

proposed an infrastructure solution

the fiber ports as the cassettes have a

fiber routing system, the Centrix system

designed around its proven Centrix™

sliding mechanism with drop-down

is a cross-functional system that meets

system.

handle.

the requirements of multiple application

solution components to site • Competitively priced, including training and consultation • A reputation for innovation with

Corning indoor/outdoor cables, The innovative design of the Centrix

typically 96 fiber, were terminated on

spaces. Telehouse was pleased with the

system enables an ultra-high-density

cassettes within the Centrix frame and

installation, as it met all requirements

deployment in a compact footprint and

installed along diverse routes to each

and was delivered on time and within

provides a scalable fiber management

of the customer suites. These cables

budget. Consultancy services and advice

solution for cross-connect applications

utilize low-loss SMF-28® Ultra optical

were provided to Telehouse throughout

in the data center’s central hub. The

fiber, which provides a solid foundation

the project process to ensure its success.

Centrix system supports up to 4,320 LC

of high-performance for the newly

connector ports per 2,200 mm frame

upgraded infrastructure. SMF-28 Ultra optical fiber offers

with a 900 mm wide, 300 mm deep

Project handover training was provided to the Telehouse operations team with further consultation and

footprint. The highest density of 17,280

industry-leading specifications for

installation support being offered on an

optical fiber ports in one square metre is

attenuation and macrobend loss. Low

ongoing basis. Telehouse is now able to

possible in a quad configuration.

attenuation enables extended reach of

provide fast and flexible provisioning

network connectivity between locations,

of connectivity to suites and respond

routing paths for jumpers, reducing

while 33 percent better macrobend

quickly to the changing business needs

the risk of entanglement, while the

performance helps improve existing

of its business customers.

operations staff can install or remove

duct utilisation and the support of

a single patch cable in less than two

smaller enclosures.

The frame design provides optimized

The project, which began in late

minutes regardless of the cable route.

A successful execution and implementation of this installation will create the basis for future project

December 2015, involved the initial

stages, including the infrastructure in a

modular cassette that can be tailored in

termination of over 16,000 fibre ports

new Telehouse building scheduled for

a variety of ways to provide flexibility

on the Centrix as well as the installation

completion later in the year.

The foundation of Centrix is a

19 | DCD eBook • datacenterdynamics.com

DCD eBook | Connectivity

Next-Gen data center connectivity: 400G and beyond As network architectures continue to grow in complexity, Cindy Ryborz marketing manager DC EMEA at Corning Optical Communications highlights the importance of keeping it simple The data center industry is

grown to reach 40G and 100G. From

What does this mean for your data

experiencing unprecedented growth

the range of connectors released to the

center?

and innovation as new players,

market within the last 30 years, there

From tenants in colocation data

new business models and new

are two interfaces, the LC duplex and

centers that pay by space, to users that

technologies converge. Despite

MPO/MTP which have made their way

are reaching capacity in their own

a slowdown in 2020, Gartner

into the standards for data centers.

facilities, achieving greater density

projects that spending on global data

However, what is fit for purpose

and preparing for future requirements

center infrastructure will reach $200

today can quickly change as the

is increasingly key. Even though today

billion in 2021 and grow year-on-year

industry moves forward and new

you may not need to implement 400G

through 2024.

architectures emerge. As data center

in your data center, increasing density

operators search for modern ways

in the main distribution area (MDA) is

support ever more rapid and seamless

to achieve faster, greater, and more

always valuable and future demands

transition of data, with technologies

powerful data rates like 400G, we

from new applications or services may

like 5G, and the applications it

must look at even higher density

consume the bandwidth available to

supports, requiring huge volumes of

connectivity and ways to simplify

date.

data to be processed, analyzed and

network design with breakout options.

MDC and SN connector formats

Beyond the readily available form

promise the possibility of connecting

Driving this growth is a need to

stored. This will see a need to not only invest in faster transceiver technology,

factors, the market is meeting this

directly from one high-speed

future-ready physical infrastructure,

need by introducing new connectors

transceiver to another transceiver,

cabling and connectivity but also

categorised as ‘Very Small Form

which simplifies the insertion of

simplify network architectures which

Factor Connectors’ (VSFFC) such as

individual connectors into various

have become more and more complex

the Senko Nano (SN) and US Conec

switches from 400G to 4x100G. In

over time.

Mini-Duplex Connector (MDC). As

addition, up to three MDC or SN

of yet, equipment manufacturers

connectors fit into the footprint of

important role in this transition and

have not released transceivers for

an LC duplex, which provides an

the past few decades have seen the

these connector interfaces, but they

enormous density advantage.

arrival of a diverse range of connector

are expected to announce their

types, some developed to support the

availability within the next couple of

reduced space in their data center,

scalability of data centers as they have

years.

implementing LC Duplex connectivity

Fiber optic connectors will play an

For operators struggling with

with LC to MDC patchcords and

What is fit for purpose today can quickly change as the industry moves forward and new architectures emerge

compatible hardware is an effective approach. This will not only allow the LC Duplex footprint to be retained at the transceiver end but the port

20 | DCD eBook • datacenterdynamics.com

density with MDC in modules or cassettes of the same size can also be increased by up to 3x – imagine the possibility of having 432 instead of 144 fibers in one rack-unit. VSFFCs also help to lower the total

For operators struggling with reduced space in their data center, implementing LC Duplex connectivity with LC to MDC patchcords and compatible hardware is an effective approach

cost of ownership. Various optical component manufacturers are already

an effect on the optical transceiver

components, as Intel demonstrated,

starting to offer solutions with these

socket that will be needed and also the

for example, with a recently

connectors, but it is important to

passive connectivity components.

published co-packaged Ethernet switch. By doing this, co-packaged

find the best cabling infrastructure solution that also allows for continued

Moving beyond pluggable optics

optics promise to increase density,

use or reuse of existing components.

The future of this industry will see a

reduce latency, reduce the power

This in turn helps to minimize the

variety of technological advances that

consumption and reduce the size of

initial investment while meeting

will help us achieve higher data rates

the switches.

future scalability requirements.

– from 40G to 100G to 400G all the way up to 1.6T.

Beyond the SN and MDC, the CS Corning-Senko Duplex (CS) represents

The use of pluggable optics may

Reaching these data rates will mean going above and beyond the current small form factor connectors

another option in the VSFF space.

play an important role up to 800G,

previously mentioned. Further

It’s important to note that, while the

and it is to be expected that certain

developments can be expected in

three types all contain two fibers, they

future developments will use some of

the expanded beam area, resulting in

have many differences in design and

the form factors already mentioned.

more applications with connectors

functionality – including size as well

However, for 1.6T, the high density

such as the US Conec MXC connector

as the vertical/horizontal orientation

and power consumption requirements

or the 3M EBO. New developments in

of the fiber.

means that pluggable optics might not

fiber such as multicore fiber (MCF) and

be the best solution.

reduced cladding fiber will also have a

The SN and MDC can also be ganged together as a 4x2 fiber

When it comes to these higher data

tangible impact on further connector

connector, which the CS cannot

rates, there is another way, namely

do. Given these variances, the CS,

the co-packaged solution. Here, data

MDC and SN receptacles are not

transmission and data processing

Preparation is key

compatible, and this therefore has

are coupled in semiconductor

The perennial challenge for data

developments.

center operators is to ensure that the networking and structured cabling design remains flexible to minimize cabling infrastructure costs when the time to upgrade to higher networking speeds like 400G and beyond arrives. Careful planning and preparation will avoid costly upgrades and changes in footprint further down the line, with increasingly compact connectors and fiber management a vital component to building 400G networks that can serve high volume telecommunications providers, enterprises and hyperscale data centers.

21 | DCD eBook • datacenterdynamics.com

DCD eBook | Connectivity

Broadcast DCD>Talks Connectivity with Jeanne Propst, Corning

Coping with requirements for increased bandwidth and port density - can this be done sustainably?

Corning: The road to 800G

22 | DCD eBook • datacenterdynamics.com