The Optical Networking and Communication
Conference & Exhibition

Los Angeles Convention Center,
Los Angeles, California, USA

Data Center and Enterprise Networks - Data Rate and Bandwidth Requirements

Data Center and Enterprise Networks - Data Rate and Bandwidth Requirements

By Lisa Huff, Discerning Analytics | Posted: 15 March 2016 4:19:51 PM
  Twitter |   LinkedIn |   Facebook | 

Even though Gigabit Ethernet has been around for more than 10 years now and 10-Gigabit Ethernet for more than seven years, 10G’s adoption rate was slow. Its deployment only passed that of Gigabit three short years ago – in 2013. The reasons are really two-fold: cost and density. As a result, we will see a more fragmented data center market. We expect small and medium enterprise data centers to continue to upgrade their servers from 1G to 10G and use 40G in the aggregation layer, while large and very large may instead transition to 25G using 100G in their aggregation layer.
In the early days of 10G, the modules were too large and too power hungry to put more than a couple on a line card. The overwhelming cost factor was the yield of the optical devices—spectral width and encircled flux specifications proved too tight for most transceiver manufacturers to meet with any kind of reasonable yield.  Until rather recently, it was still much cheaper to put 10, 1-Gigabit Ethernet devices on a card than it was to put one 10-Gigabit Ethernet device there. But, those 1G connections are much harder to manage than the one 10G one. Nonetheless, at last, due to the growth in server farms (in data centers) and the arrival of the smaller form factor SFP+, 10-Gigabit Ethernet has made considerable headway even in medium-sized companies.
With the success of 10G deployment in data center servers, we have seen an uptake of 40G for aggregation and core switches. Based on the information gathered through observations and interviews with more than 100 data centers, the following roadmap was developed.

Data Rate Roadmap (Gb/s)
  Data Center LAN
Year EDA MDA HDA/ZDA Horizontal TR WAO WAP  
2009 1/10 1/10 1/10 1 1 1 1  
2012 1/10 10/40 1/10 1/10 1 1 1  
2015 1/10 10/40/100 1/10/40 1/10/40 1/10 1 1/2.5  
Beyond 2019 10/25/40 40/100 10/25/40/100 10/40 10/40/100 1/10 1/2.5/5  
EDA = Equipment Distribution Area MDA    = Main Distribution Area
HDA = Horizontal Distribution Area ZDA      = Zone Distribution Area
TR = Telecommunications Room (Wiring Closet) WAO = Work Area Outlet  
WAP    = Wireless Access Point  
In addition to progressing to the next higher data rate of Ethernet, the IEEE has decided to also focus on some new lower speed variants to facilitate lower cost solutions for smaller enterprises and support devices like wireless access points and VoIP phones, both of which do not need 10G connections. Those variants include 2.5G and 5G as shown in the above table.
The most relevant fact about optical networking in data centers is that as these higher data rates are adopted, they are typically optical-based first. For instance, 10GBASE-SR (short-reach optical Ethernet) was adopted in the aggregation/distribution (HDA/ZDA) layer of the data center network long before the copper variants. The same happened at 40G. We do expect an exception to this at 25G, however. We expect the EDA connection (top-or-rack switch-to-server) 25G connection to be adopted as primarily copper first. But its relevance to optical solutions is clear – it will push 100G to the aggregation layer of the data center network.

Lisa Huff is Principal Analyst at Discerning Analytics. She is a Certified Data Center Manager and electrical engineer focused on market research and analysis of data center technologies. You can see some of her work at


Posted: 15 March 2016 by Lisa Huff, Discerning Analytics | with 0 comments

Blog post currently doesn't have any comments.

The views expressed in this blog are those of the authors and do not necessarily reflect the views or policies of The Optical Fiber Communication Conference and Exposition (OFC)  or its sponsors.

Sponsored by: