"Data Center to Data Center" Traffic is Growing Faster Than "Intra-Center" or "Data Center to End User" Traffic

Once upon a time, global wide area network traffic was relatively limited, and mostly represented connections between tandem switch and central office switch locations.

These days, traffic tends to be dominated by server-to-server traffic, within data centers, between data centers and then from data centers to end users.

A recent Bell Labs study forecasts total metro traffic will increase 560 percent by 2017, largely driven by IP video and the increasing adoption of cloud and data center services and applications.

According to the study, metro video traffic (including subscription TV and Internet video) will increase 720 percent by about 2017.

Metro cloud and data center traffic will increase 440 percent by 2017, the study predicts.

As the demand for video content increases, video caching is now being implemented within metro networks, moving content caching deeper into the network.

As a direct consequence, traffic between data centers in metro areas will grow, keeping much traffic off the backbone networks. That’s a significant change.

On the other hand, a growing percentage of traffic also will be moving between data centers.

Traffic between data centers is growing faster than either traffic to end-users or traffic within the data center, and by 2018,  traffic between data centers will account for almost nine percent of total data center traffic, up from nearly seven percent at the end of 2013, according to Cisco.

Until recently, metro traffic had a “north-south” flow from a content source to the end user with content sources typically located at a national central location and delivered over the wide area backbone network.

But there is a change coming, Alcatel-Lucent says. The north-south flows increasingly will be replaced by “east-west traffic  flows for traffic flows from data center to data center, increasingly located within metro centers.

There are revenue implications for providers of high-capacity metro networks and long haul bandwidth providers alike.

Data Center Traffic Now Equals Internet Traffic

For most of the last 10 years, data center to data center traffic has been a huge part of demand for wide area network transport capacity. With heavy consumer demand for web apps of all types, including streaming video, you would expect IP networks that support consumer access to carry a significant amount of traffic. 

In 2021, for example, internet traffic overall, which includes business-to-consumer and business-to-business traffic, will be roughly equivalent in magnitude. That is part of a trend that has been in place for nearly a decade, where data center to data center traffic has grown as a percentage of total traffic flowing over wide area networks. 

source: Cisco 

The implications for suppliers of WAN connectivity are significant. It now is possible to capture any demand related to public connectivity revenue simply by focusing on data center to data center connectivity. 

In other words, trends in WAN traffic and value have come to resemble the general pattern of global telecommunications, where enterprise or business demand is about equivalent to consumer demand. That does not necessarily correspond directly to revenue shares, but there is a correspondence. 

It also is possible to illustrate the value of interconnection (network effect) by examining data flowing between organizations and servers within a data center. Much of that traffic represents interconnections and data flow between collocated entities within any single data center. 

source: Cisco 

Pre-internet, connectivity providers were the main actors in collocation activity. These days computing as a service suppliers and application providers are a much bigger factor. Connectivity providers might still represent about 64 percent of interconnections, but enterprises represent 34 percent.

source: Equinix Global Interconnection Index

As suggested by the Equinix Global Interconnection Index, private interconnection happens routinely between enterprises, network, cloud and other information technology providers. 

source: Equinix

"Data Center to Data Center" Traffic Drive About Half of WAN Demand

Data center traffic moving to end users was a decade ago a larger percentage of total wide area network data volume. That has been steadily changing, with more traffic moving between data center locations.  

In 2021, the volume of data moving between data centers is about equal to the amount of data moving to end users. Content caching accounts for some of the data center to data center increase. Content mirroring accounts for an additional amount of inter-data-center traffic. 

source: Cisco 

The huge amount of “within the data center traffic” is partly caused by applications that involve lots of queries. Many internet applications are extremely “chatty”. A single search query within the data center might involve hundreds of server requests, for example. 

A social networking transaction has a similar multiplier effect, as it draws in an entire social graph to respond to a single query. 

The architecture of data centers can contribute to the amount of traffic as well, using with separate storage arrays, development or  production server pods and application server clusters that all need to talk to one another.

Still, wide area network bandwidth now is about equally composed of traffic heading for end users and traffic moving between data centers, a trend itself driven by the dominance of content as a driver of network capacity. 

source: Telegeography 

Content drives as much as 83 percent of transAtlantic traffic and 66 percent of transPacific traffic, for example. 

source: Telegeography

What Drives Global Bandwidth Demand? Everything

No single driver accounts for the strong growth of Internet bandwidth demand in every region of the world. Instead, almost every trend in communications, computing and entertainment now collectively boosts demand for capacity.

More users, more users on mobile, more people watching video, more content consumption, stored in more places and assembled from more sources, plus globalization all contribute to higher usage.

But usage alone is not the whole story. “Where” usage occurs also has changed, with the greatest volume growth in Asia. That, in turn, leads to more traffic across Asia to and from Europe, and to and from North America.

Use of more-global internet exchange points (IXPs) likewise has affected traffic flows.

Historically, most traffic transited through US-based network access points (NAPs) in which backbones interconnected their networks and exchanged traffic. As all ISPs had to connect to the US for international transit, they also used these links to exchange domestic and regional traffic, a process referred to as tromboning.

Today, regional IXPs are used to exchange more local traffic. Still, the largest number and percentages of IXPs are located in North America, Europe and Asia. That in turn drives traffic between the three regions.

By 2018, perhaps 78 percent of all workloads will be processed in cloud data centers, Cisco argues.

But traffic between data centers is growing faster than either traffic to end-users or traffic within the data center, and by 2018, traffic between data centers will account for almost nine percent of total data center traffic, up from nearly seven percent at the end of 2013. 

The high growth of this segment is due to the increasing prevalence of content distribution networks, the proliferation of cloud services and the need to shuttle data between clouds, and the growing volume of data that needs to be replicated across data centers.

Data center traffic on a global scale will grow at a 23-percent CAGR, but cloud data center traffic will grow at a faster rate (32 percent CAGR) or 3.9-fold growth from 2013 to 2018, Cisco predicts. .

Cloud workloads are expected to nearly triple (grow 2.9-fold) from 2013 to 2018, whereas traditional data center workloads are expected to see a global decline, for the first time, at a negative two-percent CAGR from 2013 to 2018.

From 2013 to 2018, the Middle East and Africa is expected to have the highest cloud traffic growth rate (54‑percent CAGR), followed by Central and Eastern Europe (39-percent CAGR) and Asia Pacific (37‑percent CAGR).

      Cloud Traffic Growth by Region, in Exabytes

Region	2013	2014	2015	2016	2017	2018	CAGR 2013–18
Asia Pacific	489	716	1,010	1,368	1,802	2,331	37%
Central and Eastern Europe	85	120	170	238	331	442	39%
Latin America	89	130	180	240	312	394	35%
Middle East and Africa	31	53	86	132	193	262	54%
North America	643	857	1,102	1,384	1,701	2,077	26%
Western Europe	311	401	502	631	791	988	26%

Source: Cisco Analysis, 2014

      Regional Distribution of Total Data Center Workloads, in Millions

Total Data Center Workloads in Millions
	2013	2014	2015	2016	2017	2018	CAGR 2013–2018
Asia Pacific	16.3	20.9	28.4	37.9	48.0	61.2	30%
Central and Eastern Europe	2.3	2.7	3.1	3.6	4.3	5.1	17%
Latin America	2.6	3.2	3.9	4.7	5.7	6.9	21%
Middle East and Africa	1.8	2.3	2.9	3.5	4.3	5.2	24%
North America	56.1	62.8	68.7	73.9	80.3	88.0	9%
Western Europe	29.2	33.3	36.5	39.7	42.2	45.1	9%

Source: Cisco Analysis, 2014

The portion of traffic residing within the data center will decline slightly over the forecast period, accounting for nearly 77 percent of data center traffic in 2013 and about 75 percent by 2018. 

Despite the decline, the majority of traffic remains within the data center because of factors such as the functional separation of application servers, storage, and databases, which generates replication, backup, and read and write traffic traversing the data center. 

Furthermore, parallel processing divides tasks and sends them to multiple servers, contributing to internal data center traffic.

More Computation, Not Data Center Energy Consumption is the Real Issue

Many observers raise key concerns about power consumption of data centers in the era of artificial intelligence. 

According to a study by the Lawrence Berkeley National Laboratory, AI-driven data center electricity consumption could increase by 50 percent to 200 percent by 2040, posing new challenges for data center operators trying to limit and reduce carbon emissions and electrical consumption. 

Study	Year Published	AI-driven electricity consumption (GWh)	Increase over 2023 (%)
Lawrence Berkeley National Laboratory	2020	130	40%
Gartner	2021	200	50%
IDC	2022	300	75%
DigiCapital	2023	400	100%
Lawrence Berkeley National Laboratory	2018	10% of total data center electricity consumption	50%
Gartner	2020	15% of total data center electricity consumption	75%
IDC	2021	20% of total data center electricity consumption	100%

Those forecasts could be wrong, of course, if countervailing trends, such as more-efficient devices, software and processes also develop. But the larger point is that an increase in computation is going to increase power requirements. 

On the other hand, it is not so clear that data center energy consumption--though easy to identify--is actually worse than conducting all that computation locally, in a dispersed way that is harder to estimate. 

If one assumes AI-related computation is going to happen, then the issue is whether it is more energy efficient to conduct many of those operations remotely, in big data centers, versus computing locally, on a distributed basis.

And there the issue is more complicated. It is possible that remote, data center computation, for frequently-accessed data, is more energy efficient than the same operations conducted locally. 

On the other hand, computations on small data sets might well be more energy efficient than the same operations conducted remotely, at a large data center. 

Study Title	Authors/Publisher	Year	Key Findings
The Energy Consumption of Cloud Storage: Exploring the Trade-Offs	Zhiwei Xu et al.	2018	Cloud storage can be more energy-efficient than local storage, especially for frequently accessed data.
The Power of Servers: A Hidden Environmental Cost of Cloud Computing	Elliot et al.	2014	Highlights the significant energy consumption of data centers but acknowledges potential efficiency gains compared to widespread local storage.
A Survey on Modeling Energy Consumption of Cloud Applications: Deconstruction, State of the Art, and Trade-Off Debates	D. Kliazovich et al.	2013	Emphasizes the importance of considering network energy consumption when comparing local vs. remote storage
How Green is the Cloud? A Comparison of the Environmental Footprint of Cloud Computing and On-Premises Solutions	M. A. van den Belt et al.	2013	Concludes that cloud storage can be more environmentally friendly for large datasets due to economies of scale and potential for renewable energy use in data centers.
Energy Consumption of Cloud Storage: The Importance of Power Management	Zhiwei Cao et al.	2011	Concludes that cloud storage can be more energy-efficient than local storage, especially for large datasets.
A Survey on Modeling Energy Consumption of Cloud Applications: Deconstruction, State of the Art, and Trade-Off Debates	George Kousiouris et al.	2018	Highlights the importance of network energy consumption when considering cloud storage. Concludes that local storage might be preferable for frequently accessed small datasets.
The Energy Efficiency of Cloud Storage Compared to Local Storage	Aapo Ristola et al.	2017	Finds that cloud storage can be more energy-efficient for most use cases, especially with increasing data volume.

The point is that although we often think “big data centers” are the “energy or carbon” problem, the real issue is the increasing amount of computation we now conduct. It is not so clear that the data centers are the real issue.

Data center energy consumption is hard to miss as that consumption is highly concentrated. Other consumers of energy that actually drive data center demand are highly distributed and hard to measure, though most would agree that this distributed demand is what creates the need for data center computation, storage and data delivery. 

Device Category	Consumer TWh	Business TWh	Total TWh	Source
Laptops & Desktops	1,200	400	1,600	The Shift Project: https://theshiftproject.org/en/home/ (2019)
Smartphones & Tablets	800	100	900	International Energy Agency (IEA): https://www.iea.org/reports/energy-efficiency-2023 (2023)
Servers (excluding data centers)	-	200	200	The Shift Project: https://theshiftproject.org/en/home/ (2019)
Network Equipment	200	100	300	The Shift Project: https://theshiftproject.org/en/home/ (2019)
TVs & Streaming Devices	600	100	700	IEA: https://www.iea.org/reports/energy-efficiency-2023 (2023)
Gaming Consoles	200	50	250	The Shift Project: https://theshiftproject.org/en/home/ (2019)
Other Devices (printers, wearables, etc.)	100	50	150	Estimated based on IEA report on standby power