Design Article
Tell us What You Think
We want to know what you thought about this Design. Let us know by adding a comment.
2012--A Return to Normalcy and Pragmatic, Power Conscious 100G
Neal Neslusan, MultiPhy
12/15/2011 6:48 AM EST
In 2010 and 2011, the industry saw the first real rollouts of 100G transport solutions. Many of these solutions were based on initial Coherent Detection transmission implementations and FPGA-based Framers. The deployment of these solutions was not unexpected and falls in line with the historical nature of the optical telecommunications marketplace; that is, in the initial years of a technology’s life the vast majority of products come from vertically integrated development teams and the focus of the product is on performance and just making it work.
In 2012 we’ll start to see 100G taking a bigger place in the build out of new and existing networks around the world. As 100G becomes more widely deployed in the larger datacenters, the push to transport 100G across the wide-area optical network will increase. This push will not only affect the largest telecom equipment makers (those with the vertical integration capabilities), but will also affect the second tier of equipment suppliers whose place in the broader network is just as significant as the largest players.
This second tier of equipment suppliers may not possess the capability to realize the complex functions of framing and module development on their own. As such, 2012 will usher in a return to normalcy with the subsystem suppliers and ASSP providers taking a more significant role in providing system solutions to the optical transport marketplace.
While this may not be a bold prediction (indeed, one need only look at the past when 10G was rolled out to properly predict the future), the timing of it is significant. In 2012, the industry will see the first true announcements and shipments of line-side 100G optical modules and 100G framer ASSPs. While this will initially affect the second tier optical system providers, the past has shown that once these technologies become readily available in the marketplace, even the largest and most vertically integrated suppliers will consider using these off-the-shelf solutions.
If we look at the advantages of these solutions we will see a deeper and more significant reason for their deployment. Certainly the prospect of not having to continually reinvest in a technology that is available on the open market is an important one to the system provider, but when we consider that many of the commercially available solutions in both the framer and optical module space will be significantly lower in power consumption than the first generation solutions, we will see an important reason for their adoption.
The year 2012 will be characterized by a sober and practical analysis of new solutions with an eye on the problem that never goes away - power consumption. If we look at many of the 100G solutions that are in the marketplace today we find significant power consumption in both the framers and the line side optical modules. In many cases the framer function is dissipating over 50 watts and the module solution is dissipating near 100 watts (with rumors of some solutions being over 100 watts). In this case nearly all the initial line side module solutions are 100G Coherent Detection solutions. When considering the adoption rate of a new technology, the industry always compares the new technology with the older technology on cost and power. Clearly there are cases in some of the largest and most dense networks where deployment of 100G transport is absolutely a must, but in other networks the deployment of 100G could be considered a luxury over the next few years if the power and cost assumptions don’t come down.
When compared to 10G, the initial deployments of 100G are clearly too costly and too power hungry to be widely deployed as the primary transport technology. While the costs vary, and are dominated in the module space by a high initial cost of low volume Coherent Detection componentry, the power multiples of 100G as compared to 10G are too great. We can assume an average power dissipation of about 3 watts for a 10G OTN Framer and 3.5 W for a 10G Tunable WDM XFP. For the framer case, 3W per 10G of bandwidth as compared to 5W per 10G of bandwidth (for 100G transmission) is not so out of line that the framer portion of the 100G deployment cannot be considered acceptable. Indeed, the initial ASSPs in the framer space need only get down to the 40W area or a little below to effectively be on par with the 10G framer power dissipation.
On the module side, however, the story is quite different, and for this consideration we need to look specifically at Coherent Detection transmission. Without even considering the issue of pluggable versus non-pluggable formats (XFP for 10G is pluggable while the 168pin MSA for 100G is not), the per-10G of bandwidth power dissipation for 100G Coherent Detection is 3X that of 10G Direct Detection. That means that while the average WDM link for 10G is dissipating about 3.5W per optical module, the average WDM link per 100G is dissipating about 3 times that or about 10W per 10G of bandwidth (about 100W total). This is a very significant difference that cannot be overcome easily with just a next generation solution. Optical technologies need to achieve greater levels of integration and silicon technologies need to jump generations in order to significantly reduce this power consumption. As such, per bit power consumption parity for 100G versus 10G (as viewed from the Coherent Detection world) is years away.
In addition, the task of cooling a 100 Watt optical module on the edge of a board is not without its challenges. Placing more than one of these modules on a single MSTP board may be impossible from a power dissipation and cooling perspective. And many of the newest generation MSTP architectures support more than 100G per backplane slot, making the desire to realize more than one 100G optical transponder per board a real one.
The result is that in the year 2012 the focus of the optical transport marketplace will move to much lower power and lower cost Direct Detection optical transport solutions. Year 2012 will see significant interest in both 4X28G and 10X10G Direct Detection transport schemes as their power footprints are more in line with what 10G offers them today. Clearly the spectral density of these solutions does not match that of 100G Coherent but in many cases, perhaps in most cases, spectral density is not the primary concern of the network. Transporting a single 100G OTU4, and the management cost savings of that single signal over transporting 10 OTU2 signals, will drive the move to these technologies. And many of the boxes that will need to transport this signal simply cannot withstand the power dissipation of 100 Watts for a 100G optical transponder.
While 2012 will see a reassertion of the traditional suppliers of optical transport technologies, the solutions that come from these suppliers will be measured by the same old traditional criteria of cost and power. But in this generation of system development, power consumption may ultimately be the final deciding factor. The result of this will be that while Coherent Detection will be deployed in 2012 for 100G by many vendors in many networks around the world, its widespread and ubiquitous deployment may be years away as system vendors eschew it in favor of lower cost, lower power Direct Detection technologies.
About the Author
Neal Neslusan is a VP of Sales and Marketing at MultiPhy.
In 2012 we’ll start to see 100G taking a bigger place in the build out of new and existing networks around the world. As 100G becomes more widely deployed in the larger datacenters, the push to transport 100G across the wide-area optical network will increase. This push will not only affect the largest telecom equipment makers (those with the vertical integration capabilities), but will also affect the second tier of equipment suppliers whose place in the broader network is just as significant as the largest players.
This second tier of equipment suppliers may not possess the capability to realize the complex functions of framing and module development on their own. As such, 2012 will usher in a return to normalcy with the subsystem suppliers and ASSP providers taking a more significant role in providing system solutions to the optical transport marketplace.
While this may not be a bold prediction (indeed, one need only look at the past when 10G was rolled out to properly predict the future), the timing of it is significant. In 2012, the industry will see the first true announcements and shipments of line-side 100G optical modules and 100G framer ASSPs. While this will initially affect the second tier optical system providers, the past has shown that once these technologies become readily available in the marketplace, even the largest and most vertically integrated suppliers will consider using these off-the-shelf solutions.
If we look at the advantages of these solutions we will see a deeper and more significant reason for their deployment. Certainly the prospect of not having to continually reinvest in a technology that is available on the open market is an important one to the system provider, but when we consider that many of the commercially available solutions in both the framer and optical module space will be significantly lower in power consumption than the first generation solutions, we will see an important reason for their adoption.
The year 2012 will be characterized by a sober and practical analysis of new solutions with an eye on the problem that never goes away - power consumption. If we look at many of the 100G solutions that are in the marketplace today we find significant power consumption in both the framers and the line side optical modules. In many cases the framer function is dissipating over 50 watts and the module solution is dissipating near 100 watts (with rumors of some solutions being over 100 watts). In this case nearly all the initial line side module solutions are 100G Coherent Detection solutions. When considering the adoption rate of a new technology, the industry always compares the new technology with the older technology on cost and power. Clearly there are cases in some of the largest and most dense networks where deployment of 100G transport is absolutely a must, but in other networks the deployment of 100G could be considered a luxury over the next few years if the power and cost assumptions don’t come down.
When compared to 10G, the initial deployments of 100G are clearly too costly and too power hungry to be widely deployed as the primary transport technology. While the costs vary, and are dominated in the module space by a high initial cost of low volume Coherent Detection componentry, the power multiples of 100G as compared to 10G are too great. We can assume an average power dissipation of about 3 watts for a 10G OTN Framer and 3.5 W for a 10G Tunable WDM XFP. For the framer case, 3W per 10G of bandwidth as compared to 5W per 10G of bandwidth (for 100G transmission) is not so out of line that the framer portion of the 100G deployment cannot be considered acceptable. Indeed, the initial ASSPs in the framer space need only get down to the 40W area or a little below to effectively be on par with the 10G framer power dissipation.
On the module side, however, the story is quite different, and for this consideration we need to look specifically at Coherent Detection transmission. Without even considering the issue of pluggable versus non-pluggable formats (XFP for 10G is pluggable while the 168pin MSA for 100G is not), the per-10G of bandwidth power dissipation for 100G Coherent Detection is 3X that of 10G Direct Detection. That means that while the average WDM link for 10G is dissipating about 3.5W per optical module, the average WDM link per 100G is dissipating about 3 times that or about 10W per 10G of bandwidth (about 100W total). This is a very significant difference that cannot be overcome easily with just a next generation solution. Optical technologies need to achieve greater levels of integration and silicon technologies need to jump generations in order to significantly reduce this power consumption. As such, per bit power consumption parity for 100G versus 10G (as viewed from the Coherent Detection world) is years away.
In addition, the task of cooling a 100 Watt optical module on the edge of a board is not without its challenges. Placing more than one of these modules on a single MSTP board may be impossible from a power dissipation and cooling perspective. And many of the newest generation MSTP architectures support more than 100G per backplane slot, making the desire to realize more than one 100G optical transponder per board a real one.
The result is that in the year 2012 the focus of the optical transport marketplace will move to much lower power and lower cost Direct Detection optical transport solutions. Year 2012 will see significant interest in both 4X28G and 10X10G Direct Detection transport schemes as their power footprints are more in line with what 10G offers them today. Clearly the spectral density of these solutions does not match that of 100G Coherent but in many cases, perhaps in most cases, spectral density is not the primary concern of the network. Transporting a single 100G OTU4, and the management cost savings of that single signal over transporting 10 OTU2 signals, will drive the move to these technologies. And many of the boxes that will need to transport this signal simply cannot withstand the power dissipation of 100 Watts for a 100G optical transponder.
While 2012 will see a reassertion of the traditional suppliers of optical transport technologies, the solutions that come from these suppliers will be measured by the same old traditional criteria of cost and power. But in this generation of system development, power consumption may ultimately be the final deciding factor. The result of this will be that while Coherent Detection will be deployed in 2012 for 100G by many vendors in many networks around the world, its widespread and ubiquitous deployment may be years away as system vendors eschew it in favor of lower cost, lower power Direct Detection technologies.
About the Author
Neal Neslusan is a VP of Sales and Marketing at MultiPhy.
|
|
|
|
|
Navigate to related information
