| Thread Links | Date Links | ||||
|---|---|---|---|---|---|
| Thread Prev | Thread Next | Thread Index | Date Prev | Date Next | Date Index |
|
Scott
Thank you for digging up this data. However, it is a shame you didn’t actually listen to my presentation and rather dead set on proclaiming my data and information is far fetched and a lie.
So let me take my presentation point by point …
The point of my presentation was to establish a line of reasoning supporting a 3m no FEC case for the 25GE effort.
Length
I spent considerable time looking at ways data center, service provider / Core, and Enterprise would implement a 2m no FEC solution. We want a solution, in my opinion, that satisfies a broad range of the market. There are many cases in the enterprise
and core markets that 2m won’t satisfy. I listed those examples. I think there has been significant data showing mid front to top front, and mid front to bottom front. I went through the lengths, discussed the rack, the hood system, and the dynamic around
the use case.
FEC Latency
There is a segment of the market that does not want to add unwanted time into the connections. Further, it was discussed that there might be several levels, certainly two levels, such that latency begins to add up.
FEC power
You show 2 numbers that are 50% apart. There is obviously differences in FEC implementations and there are designs that are highly optimized and those that are not.
In my presentation, I specifically said I wanted to avoid an argument on “my FEC power is better then yours”, so I used a non optimized design case recommended to me. My intent was to demonstrate that FEC does add power and that this does add up. Data
centers, HPC, cloud computing, and core streaming have limited power. Saving any amount of power allows these business models to add more servers or more streaming devices, regardless of the cost. The point I was making is there there is no low hanging fruit
left to save power. We now need to look at making a lot of little changes to sum up to the best savings possible. Eliminating FEC helps.
Further, lets examine how much power actually is used for that power. There are, in north america, one or two transformer steps. Each has a penalty of 5%-7% loss. Then there is the power conversion at the supply and again at the point of load, which
for a range of equipment can be a loss of 20% to as low as 12%, depending on conversion steps and efficiency. Then there is cooling power, which can account for 12% to 47%, depending on implementation and RU height. There is also the facility cooling loss.
Most large facilities target PUE of 1, but often can not get there. Google, MicroSoft, FaceBook are all doing great. But they are not the only place in the market that buys this stuff. Google also buys and sells energy, not true for others. Your use
of wholesale is not exactly true. Google owns wind power in Iowa, which they use to offset costs in the west coast. Others are in Washington where hydro power is really cheap. As I mentioned to you, electricity costs for corporations range from about 5
cents to 33 cents, DOE numbers, throughout the world.
I think you are grossly unaware of the point of 3m no FEC … which is that there is a need for low latency, power efficient, 3 meter solutions. Thanks for sharing your data, but it isn’t really useful for this discussion.
Joel
From: Scott Kipp <skipp@xxxxxxxxxxx>
Reply-To: Scott Kipp <skipp@xxxxxxxxxxx> Date: Tuesday, July 14, 2015 at 4:26 PM To: "STDS-802-3-25G@xxxxxxxxxxxxxxxxx" <STDS-802-3-25G@xxxxxxxxxxxxxxxxx> Subject: [STDS-802-3-25G] Power Consumption of of 25G RS-FEC 802.3by, Yesterday, there was considerable debate about power consumption due to the RS-FEC used in 25GbE. In goergen_3by_02_0715.pdf, the power consumption for RS-FEC was approximated to be 300mW. I did a quick search
and found these estimates: http://www.ieee802.org/3/bm/public/sep12/wang_01_0912_optx.pdf = 45mW on slide 3 http://www.ieee802.org/3/bj/public/mar12/gustlin_01_0312.pdf = 90 mW on page 6 http://www.ieee802.org/3/bs/public/adhoc/logic/oct21_14/wangz_01_1014_logic.pdf, Wangz concludes: In practice, either
KR4 or KP4 FEC is easy to implement and not much power-consuming. A conservative estimate is 100mW for 100GBASE-KR4. 25GBASE-KR4 will consume 1/4 of this power or 25mW based on a one of the four lanes. I propose that 25mW is a practical power consumption for RS-FEC. Let’s calculate the cost of the power consumption for the server over a 3-year lifespan.
Another correction that I wanted to make is to the estimate is the cost of electricity. Hyperscale data centers that consume megawatts of power buy electricity at wholesale rates. Many are located near electric
power plants so that Google pays about $0.04/kWh according to this article: http://www.oregonlive.com/silicon-forest/index.ssf/2013/09/google_reaches_new_data_center.html The wholesale electricity industry is regulated and current pricing is available for this site and shows electricity ranges in 2015 has varied from $0.036/MWh to $0.065/MWh. Read more here: http://www.eia.gov/electricity/wholesale/ I used a conservative $0.05/kWh. My estimate is that the cost of RS-FEC per server is about $0.03 over the life of the server. Three cents is basically in the noise for our purposes and should not be a fundamental driver for 3m cabling without
FEC. Kind regards, Scott Kipp |